merge: hotfix/fresh-box-experience → main (v0.3.5 fresh-box experience)

Fixes the v0.3.4 fresh-box failures: silent CLI under bare-word PATH
invocation (SEA argv detection), bootstrap dying on venv creation without an
actionable hint (poisoned-partial-venv recovery + apt hint + doctor
venv-capability probe + preflight), installer falsely 'verifying' a silent
binary, and localhost-only server binding (network mode: 0.0.0.0 + wildcard
CORS + hostname-derived API URL — remote browsing zero-config).

38 CLI tests, 3 web tests, 409 ai-service tests green; build/typecheck/lint
clean.

---ci---
phase: hotfix
milestone: v0.4
status: complete
type: hotfix
requirements:
  covered: [REQ-4-001, REQ-4-002, REQ-4-003, REQ-4-004, REQ-4-005]
  partial: []
---/ci---

.ciagent/ARCHITECTURE.md

@@ -6,6 +6,8 @@ Nextcraft is a TypeScript monorepo (pnpm workspaces + turborepo) with a Next.js
 
 **v0.3 additions (Credential Engines):** real credential engines replace v0.2 mock inputs — a sandbox fabric (isolated per-learner coding environments via Linux user/mount/pid/net namespaces), a live build-telemetry pipeline (WebSocket ingest + SQLite-ordered event log), a process-trace grading engine, seeded per-learner variant task generation, and a voice-based oral defense (STT/TTS via a new provider-agnostic voice layer). **First real persistence introduced: SQLite** (`ai_service/telemetry/`, grading, variant, defense stores). Lab/Assessor/Proctor agents are re-grounded onto real telemetry/traces. **Identity/age-gating (KYC) deferred per founder directive** — no security engineer persona; secrets-hygiene checklist only.
 
+**v0.4 additions (Distribution & Bootstrap CLI, founder directive D-016):** a new `apps/cli` package — the `nextcraft` bootstrap CLI (`doctor`/`bootstrap`/`verify`/`dev`) compiled to a self-contained linux x64 binary via **Node SEA** (probe-verified: Go/Rust absent, node v24.15.0 SEA-capable), installed by a repo-served one-liner script that resolves the latest Gitea release, downloads binary + sha256 sidecar, verifies, and installs to `~/.local/bin`. Every release from v0.4 onward attaches the binary + checksum as release assets (the "ongoing binaries" requirement). The CLI is a thin wrapper: all orchestration logic stays in `apps/ai-service/scripts/` (bootstrap.sh/dev.sh) — the CLI composes them via subprocess (A-202), duplicating nothing. Previously-planned v0.4 seams (real STT/TTS, KYC, design/sim envs, seq-lease) move to v0.5.
+
 ### Confirmed Technology Stack (v0.2)
 
 | Technology | Version | Purpose |
@@ -26,6 +28,8 @@ Nextcraft is a TypeScript monorepo (pnpm workspaces + turborepo) with a Next.js
 | pydantic | 2.13.x | Request/response models, structured outputs |
 | pydantic-settings | 2.15.x | Settings + env-file loading (replaces python-dotenv) |
 | httpx | 0.28.x | Async LLM HTTP client (ollama-cloud + local providers) |
+| sqlmodel / sqlalchemy | 0.0.24 / 2.x | Typed SQLite persistence for the v0.3 engine stores (D-027) |
+| python-multipart | 0.0.x | Multipart audio upload for the defense answer route (REQ-3-006) |
 | sse-starlette | 3.4.x | SSE framing, ping keep-alive |
 | pytest | 9.x | Test runner |
 | pytest-asyncio | 1.4.x | Async tests (auto mode) |
@@ -45,6 +49,15 @@ Deliberately **not** used: openai-python SDK (the `LLMProvider` protocol is the
 7. **D-022 Monorepo integration** — zero-dependency shim `package.json` in apps/ai-service + `ai#*` turbo passthrough tasks (`cache:false, outputs:[]`) + root `ai:dev`/`ai:test` scripts + idempotent venv bootstrap.
 8. **D-023 Testing** — pytest-asyncio auto mode; TestClient `client.stream()` for SSE; httpx MockTransport for byte-exact provider parser tests; scripted mock provider incl. failure modes. Tests never call the cloud.
 
+### v0.4 Architecture Decisions (from Research — Distribution & Bootstrap CLI)
+
+18. **D-033 Binary toolchain = Node SEA (probe-verified)** — Go and Rust are absent from this box; node v24.15.0 ships SEA support (`--experimental-sea-config`, postject-free on linux via `cp node nextcraft && node sea-config` … blob injection with the system `dd`/`npx postject` if needed). CLI source lives in `apps/cli` (TypeScript, compiled to a single CJS bundle by esbuild, then SEA-injected into a copy of the node binary → `nextcraft-linux-x64`). Fallback if SEA breaks: python3 `zipapp` (3.11.2 available). No new toolchain deps beyond dev-scoped esbuild.
+19. **D-034 CLI = thin wrapper, orchestration stays in scripts/** — `nextcraft` composes `apps/ai-service/scripts/bootstrap.sh` and `scripts/dev.sh` equivalents via `spawn` with inherited stdio and timeout guards (A-202/A-209). doctor/bootstrap/verify implement only *checking* logic (prereqs, env template, health) — never re-implement installs. This keeps one source of truth for bootstrap semantics.
+20. **D-035 Install path = repo raw `install.sh` + Gitea latest-release API** — the one-liner `curl -fsSL <forge>/coreci/nextcraft/raw/main/scripts/install.sh | bash` resolves `GET /api/v1/repos/coreci/nextcraft/releases/latest`, downloads the `nextcraft-linux-x64` + `nextcraft-linux-x64.sha256` assets, verifies sha256 (`shasum -a 256`), installs to `~/.local/bin` (PATH hint), and degrades to printed source-bootstrap instructions when no binary asset exists or the platform mismatches (A-203/A-204/A-206).
+21. **D-036 Ongoing binaries = ship-workflow asset step** — the release pipeline (v0.3's `ShipWorkflow.createRelease` equivalent, executed as the ship step's asset stage) builds the binary + checksum and attaches both to every Gitea release from v0.4 onward (A-205). Token resolution stays `.env*`-only (D-006/D-014); binaries are linux x64 only for v0.4 (macOS arm64 deferred — unverifiable on this box).
+22. **D-037 CLI package layout** — `apps/cli` is a pnpm workspace package (`@nextcraft/cli`): `src/` (entry, commands/, checks/, lib/), `scripts/build-binary.mjs` (esbuild bundle → SEA inject), unit tests runnable via `pnpm --filter @nextcraft/cli test` (node:test, no new test framework). Root `package.json` gains `cli:*` passthrough scripts mirroring the `ai:*` pattern (D-022).
+23. **D-038 Network mode (v0.3.5)** — dev binds 0.0.0.0 (`AI_HOST`, default 0.0.0.0, revert via 127.0.0.1); CORS + WS-origin gates read `AI_CORS_ORIGINS` (default `*` — any origin, safe only because credentials are never enabled; explicit comma list restricts); the web client derives the API base URL from the browser hostname at runtime (`engine-base-url.ts`: `NEXT_PUBLIC_AI_SERVICE_URL` override → `http://${window.location.hostname}:8420` → `localhost` server-side). Hotfix also fixes: SEA direct-run detection (`require("node:sea").isSea()` — argv shape differs by invocation style), installer honesty gate (silent `--version` = hard fail), bootstrap venv recovery (poisoned partial `.venv` removal + distro-specific `apt install python3.XX-venv` hint), and doctor venv-capability probe with bootstrap preflight.
+
 ### v0.3 Architecture Decisions (from Research — Credential Engines)
 
 9. **D-024 Sandbox isolation = Linux namespaces via `unshare`** — per-learner sandbox runs as a subprocess entered into fresh user+mount+pid+network namespaces (`unshare --user --map-root-user --mount --pid --fork --net`). Probe-verified on this box: in-namespace uid=0, **network fully isolated** (0 interfaces), learner writes land in a per-sandbox directory; proc-remount not permitted here but not required. Chosen because no container runtime (docker/podman/bwrap/firejail) exists on the box and there is no sudo. A `SandboxBackend` protocol abstracts the spawner so a future containerd/runc backend can replace namespace-spawning without touching callers.
@@ -65,14 +78,14 @@ Deliberately **not** used: openai-python SDK (the `LLMProvider` protocol is the
 
 | Component | Description | Boundaries | Depends On |
 |-----------|-------------|------------|------------|
-| `ai_service/main.py` | FastAPI app factory, lifespan (httpx client pool, provider factory), CORS (localhost only), /health | App entry | config, llm, agents, api |
+| `ai_service/main.py` | FastAPI app factory, lifespan (httpx client pool, provider factory, SandboxManager + reaper loop, SQLite engine stores on app.state), CORS (localhost only, incl. PUT for file writes), /health | App entry | config, llm, agents, api, engines |
 | `ai_service/config.py` | pydantic-settings Settings (env_prefix="AI_", env_file, SecretStr key) | Configuration only | None |
-| `ai_service/api/` | Endpoints: chat.py (POST /v1/chat/stream, SSE), lab.py, assessment.py, proctor.py, mentor.py; deps.py (DI) | Composes agents + sessions; never imported by llm/ or agents/ | agents, llm |
+| `ai_service/api/` | Endpoints: chat.py (POST /v1/chat/stream, SSE), lab.py, assessment.py (POST /v1/assessment/evaluate v0.2 + POST /v1/assessment/grade v0.3), proctor.py, mentor.py, sandboxes.py (lifecycle + files/exec routes, G-5 abuse gates), telemetry.py (WS ingest + trace/gaps reads), variants.py (seeded per-learner variants), defense.py (defense loop, REQ-3-006); deps.py (DI) | Composes agents + sessions + engines; never imported by llm/ or agents/ | agents, llm, sandbox, telemetry, grading, variants, voice |
 | `ai_service/llm/` | types.py (Message; ChatDelta/ChoiceDelta removed in P3 — no consumers), base.py (LLMProvider protocol), openai_compat.py (ollama-cloud + local), mock.py (deterministic), factory.py | Never imports agents/ or api/ | config |
-| `ai_service/agents/` | base.py (BaseAgent ABC), registry.py, session.py (SessionStore), structured.py (JSON defense), coach/tutor/lab/assessor/proctor/mentor.py | Never imports api/ | llm, prompts, corpus |
+| `ai_service/agents/` | base.py (BaseAgent ABC), registry.py, session.py (SessionStore), structured.py (JSON defense), coach/tutor/lab/assessor/proctor/mentor.py + examiner.py (seventh agent, v0.3) | Never imports api/ | llm, prompts, corpus, telemetry |
 | `ai_service/prompts/` | Per-agent system prompt constants + render_context functions (str.format_map) | Data only | None |
-| `ai_service/corpus/` | Mock engine inputs: learner_context.py, telemetry.py (Lab/Proctor scenarios), artifacts.py (pre-baked artifacts, rubrics, transcripts) | Pydantic-typed; aligned with TS packages/mock-data by convention | None |
-| `scripts/` | bootstrap.sh (venv + pip install idempotent), dev.sh (exports keys from .ciagent/.env.secrets → uvicorn), test.sh (pytest), lint.sh (ruff check, G-3) | Dev entry points | pyproject.toml |
+| `ai_service/corpus/` | Mock engine inputs: learner_context.py, telemetry.py (Lab/Proctor scenarios), artifacts.py (pre-baked artifacts, rubrics, transcripts). Since v0.3 P6 these are DORMANT, test-only fixtures (dormant-header noted) — the live learner path uses real engine inputs | Pydantic-typed; aligned with TS packages/mock-data by convention | None |
+| `scripts/` | bootstrap.sh (venv + pip install idempotent), dev.sh (exports keys from .ciagent/.env.secrets → uvicorn), test.sh (pytest), lint.sh (ruff check, v0.2 G-3) | Dev entry points | pyproject.toml |
 | `tests/` | conftest.py (mock provider, settings override, TestClient), health, llm (MockTransport parser), agents (framework + per-agent), api (SSE stream tests) | Mock provider only — no cloud | all |
 
 **Module boundary rules:** `llm/` never imports `agents/` or `api/`; `agents/` never imports `api/`; `api/` composes both via DI. `corpus/` is the only home of mock engine data. Prompts are code — versioned and reviewed in git.
@@ -85,32 +98,46 @@ Deliberately **not** used: openai-python SDK (the `LLMProvider` protocol is the
 | `ai_service/telemetry/` | `models.py` (TelemetryEvent, TraceSpan), `store.py` (TraceStore protocol + SQLite impl D-027), `ingest.py` (WebSocket /v1/telemetry/ingest, seq gap detection D-026) | Persistence; never imports agents/ | config |
 | `ai_service/grading/` | `features.py` (deterministic trace digest D-028), `engine.py` (rubric scoring orchestration), `store.py` (GradeStore) | LLM only via digest; never sees raw trace | llm, telemetry, prompts |
 | `ai_service/variants/` | `templates.py` (task template library), `generator.py` (seeded LLM instantiation D-029), `store.py` (VariantStore) | LLM via structured output | llm, grading |
-| `ai_service/voice/` | `base.py` (VoiceProvider protocol D-030), `openai_audio.py` (STT/TTS vs compatible endpoint), `browser.py` (native SR/TTS fallback descriptor), `mock.py` (deterministic) | Never imports agents/ or api/ | config |
+| `ai_service/voice/` | `base.py` (VoiceProvider protocol D-030), `browser.py` (native SR/TTS fallback descriptor), `mock.py` (deterministic; the real server STT/TTS provider is the v0.4 seam — GRILL CUT-1/G-7), `factory.py` (provider selection), `defense_store.py` (DefenseStore: transcripts + integrity signals, D-027) | Never imports agents/ or api/ | config |
 | `ai_service/agents/examiner.py` | Seventh agent: oral defense examiner; streams over existing SSE, consumes process traces + emits integrity signals | reuses BaseAgent (D-018) | llm, prompts, telemetry |
 | `ai_service/data/*.db` | SQLite databases (telemetry/grades/variants/defenses) | gitignored | — |
 | `scripts/sandbox-agent.py` | Tiny in-namespace capture process shipped into the sandbox; streams telemetry to ingest | standalone | stdlib only |
 
 **Boundary additions:** `sandbox/`, `telemetry/`, `grading/`, `variants/`, `voice/` are engine modules — they never import `api/` (which composes them via DI) and never import `agents/` (agents call engines through narrow interfaces, not vice versa).
 
+### apps/cli — Nextcraft Bootstrap CLI (v0.4 NEW)
+
+| Component | Description | Boundaries | Depends On |
+|-----------|-------------|------------|------------|
+| `src/index.ts` | Entry: arg parsing (no deps beyond node stdlib at runtime), command dispatch, `--help`/`--version`, exit-code contract (0 ok / 1 failure / 2 usage) | CLI surface only | commands/ |
+| `src/commands/` | `doctor.ts` (prereq checks + actionable errors), `bootstrap.ts` (pnpm install + scripts/bootstrap.sh wrapper + env template copy + key validation), `verify.ts` (health: venv imports, ports, env, build readiness), `dev.ts` (thin passthrough to scripts/dev.sh) | Compose checks/ + lib/; spawn scripts — never re-implement them | checks/, lib/ |
+| `src/checks/` | Pure check functions: `check-command.ts` (binary-on-PATH + version compare), `check-env.ts` (template diff, required/optional key classification) | Pure logic, unit-testable, no fs side effects at import | None |
+| `src/lib/` | `spawn.ts` (subprocess with timeout + inherited stdio), `log.ts` (✓/✗/warn output formatter) | Shared utilities | None |
+| `scripts/build-binary.mjs` | esbuild → CJS bundle → Node SEA injection → `dist/nextcraft-linux-x64` + sha256 sidecar | Build-time only | esbuild (dev dep) |
+| `scripts/install.sh` | The one-liner install script served from repo raw: Gitea latest-release resolve → download + checksum verify → ~/.local/bin; source-bootstrap fallback | Standalone POSIX sh | forge API |
+| `tests/` | node:test unit tests: command dispatch, check logic, env template diff, install-script shellcheck-style assertions | Fixtures only — never mutate repo state | src/ |
+
+**Boundary rules:** the CLI never imports from `apps/web`, `packages/*`, or `ai_service` Python modules — it orchestrates them exclusively via subprocess/filesystem. Runtime deps: node stdlib only (no runtime npm deps; esbuild is dev-only). The binary embeds the bundle; `scripts/bootstrap.sh` remains the single source of bootstrap truth (D-034).
+
 ### apps/web — Next.js Application
 
 | Component | Description | Boundaries | Depends On |
 |-----------|-------------|------------|------------|
-| `app/(learner)/` | Learner surface route group: landing, catalog, competency stack, dashboard, byte viewer, sandbox mockup, assessment mockup | Learner-only routes and layouts | packages/ui, packages/mock-data, packages/types |
+| `app/(learner)/` | Learner surface route group: landing, catalog, competency stack, dashboard, byte viewer, build surface (`/build/[competencyId]` — real in-browser build), defense surface (`/defend/[competencyId]` — live oral defense + grading) | Learner-only routes and layouts | packages/ui, packages/mock-data, packages/types |
 | `app/(marketplace)/` | Marketplace surface route group: job board, job detail, employer profile, search/filter, pricing | Marketplace-only routes and layouts | packages/ui, packages/mock-data, packages/types |
 | `app/(employer)/` | Employer dashboard route group: overview, talent search, candidate profile, posting management | Employer-only routes and layouts | packages/ui, packages/mock-data, packages/types |
 | `app/(admin)/` | Admin surface route group: overview, learner management, competency graph viewer, moderation | Admin-only routes and layouts | packages/ui, packages/mock-data, packages/types |
 | `app/layout.tsx` | Root layout: theme provider, navigation shell, responsive container | All routes | packages/ui |
-| `components/` | Surface-specific components (learner/, marketplace/, employer/, admin/) plus shared chrome (navigation-shell, header/footer, role-switcher, theme-provider, breadcrumbs, dark-mode-toggle) | App-level components | packages/ui |
-| `hooks/` | use-chat-stream.ts — SSE client hook: fetch + ReadableStream, byte buffering + frame reassembly, idempotent AbortController cleanup | Client components only | ai-service SSE |
-| `lib/` | sse.ts (shared SSE frame parser — CRLF normalization + `: ping` immunity, G-1), breadcrumbs.ts, format.ts | Pure utilities | None |
+| `components/` | Surface-specific components (learner/, marketplace/, employer/, admin/) plus shared chrome (navigation-shell, header/footer, role-switcher, theme-provider, breadcrumbs, dark-mode-toggle); v0.3 learner: build-surface, sandbox-terminal (read-only exec output), defense-session | App-level components | packages/ui |
+| `hooks/` | use-chat-stream.ts — SSE client hook: fetch + ReadableStream, byte buffering + frame reassembly, idempotent AbortController cleanup; use-sandbox-session.ts (v0.3) — sandbox lifecycle for the build session: create on task open, destroy on unmount, mid-start failure cleanup, 503/403/429 honest surfaces | Client components only | ai-service SSE / engine API |
+| `lib/` | sse.ts (shared SSE frame parser — CRLF normalization + `: ping` immunity, v0.2 G-1), breadcrumbs.ts, format.ts, engine-base-url.ts (v0.3.5: runtime API base — env override → browser hostname → localhost), engine-client.ts (v0.3: typed fetch client for /v1/sandboxes, files/exec, variants, grade, defense, traces) | Pure utilities | None |
 
 ### packages/ui — Shared Component Library
 
 | Component | Description | Boundaries | Depends On |
 |-----------|-------------|------------|------------|
 | `tokens/` | Design tokens as TS constants: colors, spacing, radii, shadows, breakpoints (mirrored as Tailwind v4 `@theme` tokens in apps/web globals.css) | Foundation layer — no dependencies | None |
-| `primitives/` | Button, Input, Card, Badge, Avatar — each with a Storybook story | Atomic UI components | tokens, packages/types |
+| `primitives/` | Button, Input, Card, Badge, Avatar (v0.1) + TerminalFrame, TelemetryStatus, MicControl, GradeBadge, TranscriptViewer (v0.3 build/defense surfaces) — each with a Storybook story | Atomic UI components | tokens, packages/types |
 
 Composite/layout/theme components (navigation shell, tables, chat panels, graph viewer, theme provider) live in `apps/web/components/` as app-level components, not in packages/ui.
 
@@ -134,11 +161,42 @@ Composite/layout/theme components (navigation shell, tables, chat panels, graph
 | `marketplace.ts` | Job, Employer, Candidate, JobPosting, TalentMatch, SearchFilter | Marketplace types | None |
 | `user.ts` | Learner, Admin, EmployerUser, AgeGroup, Role | User types | None |
 | `ui.ts` | Component props, theme config, breakpoint definitions | UI types | None |
+| `telemetry.ts` | TelemetryEvent/ExecResult wire shapes for the live build surface (v0.3) | Engine types | None |
+| `variants.ts` | Variant/TaskTemplate shapes for per-learner task statements (v0.3) | Engine types | None |
+| `grading.ts` | GradeRecord/RubricScore shapes for live grading display (v0.3) | Engine types | None |
+| `defense.ts` | DefenseSession/transcript/integrity-signal shapes for the defense surface (v0.3) | Engine types | None |
 
 ---
 
 ## Data Flow
 
+### v0.3 credential flow (current)
+
+```
+[learner build surface /build/*]                [learner defense surface /defend/*]
+   file CRUD + Run/Test (HTTP)                     mic MediaRecorder / typed + TTS playback
+        │                                                  │
+        ▼                                                  ▼
+[api/sandboxes files/exec] ──exec──▶ [namespace sandbox]    [api/defense start/answer/finish]
+        │                            │ capture agent        │
+        │                            ▼ (WS telemetry)      ▼
+        │                    [api/telemetry ingest]   [DefenseStore (SQLite)]
+        │                            │ SQLite               │ transcript + integrity signals
+        │                            ▼                      │
+        │                      [TraceStore] ────▶ [GradingEngine: digest (grading/features)
+        │                            │                + rubric LLM (D-028)] ──▶ [GradeStore]
+        │                            ▼                      ▼
+        └──▶ Lab agent (live digest)          Assessor (grade output) / Proctor (integrity)
+                                             Examiner agent (SSE) ◀── defense sessions
+   variants: [api/variants] ◀── [VariantStore (seeded, D-029)] ── per-learner task statements
+```
+
+- Lab consumes the live trace digest; Assessor consumes grading-engine output; Proctor consumes telemetry + defense integrity signals (REQ-3-007) — no mock fallback in the learner path (v0.2 corpus scenarios are dormant test-only fixtures).
+- The learner's path is: variant task → in-sandbox build (telemetry streams to SQLite) → grade My Work (rubric scores from the real trace) → oral defense → verdict.
+- Flooded/gapped traces are terminal: ingest closes 1008 and marks INCOMPLETE_FLOODED (G-3); the grader returns UNGRADABLE_TRACE_INCOMPLETE (G-4) — no credential from an incomplete trace.
+
+### v0.2 chat flow (complete, still live)
+
 ```
 [packages/mock-data + packages/types]      [ai_service/corpus]
         │ (TS, web surfaces)                      │ (Python, agent inputs)
@@ -153,21 +211,27 @@ Composite/layout/theme components (navigation shell, tables, chat panels, graph
                                                (https://ollama.com/v1)
 ```
 
-- Web surfaces remain server-component-first; client components (chat, filters, graph viewer, dark mode toggle) fetch directly from ai-service over SSE (A-002: no Next.js API-route proxy in v0.2).
-- The LLM provider layer is a dumb pipe — OpenAI-compatible chunks pass through byte-identical; envelope logic (meta/done/error) lives only in the API layer (D-016).
-- Lab/Assessor/Proctor read mock scenarios from `ai_service/corpus/` — real engines are v0.3+.
+- Web surfaces remain server-component-first; client components (chat, filters, graph viewer, dark mode toggle) fetch directly from ai-service over SSE (A-002: no Next.js API-route proxy).
+- The LLM provider layer is a dumb pipe — OpenAI-compatible chunks pass through byte-identically; envelope logic (meta/done/error) lives only in the API layer (D-016).
+- The v0.2 corpus scenarios (`ai_service/corpus/`) are retained as dormant, test-only fixtures (dormant-header noted); they are no longer inputs to the live learner path.
 - All automated tests use the deterministic mock provider; the cloud is for manual probes only.
 
 ---
 
-## Build Order (v0.3)
+## Build Order (v0.4)
+
+1. **Bootstrap CLI core** — apps/cli package: doctor checks (node/pnpm/python3/git/unshare), bootstrap wrapper (pnpm install + scripts/bootstrap.sh + .env template + key validation), verify health check, dev passthrough; unit tests
+2. **Binary build + release pipeline** — esbuild bundle → Node SEA binary (`nextcraft-linux-x64`) + sha256 sidecar; install.sh one-liner (Gitea latest-release resolve + checksum verify + PATH install); release-asset upload wired into the ship flow (ongoing binaries from v0.4 onward)
+3. **Install docs + fresh-clone E2E** — README quickstart (one-liner → doctor → bootstrap → dev), CLI reference, fresh-clone end-to-end test proving a clean clone reaches a running stack
+
+## Build Order (v0.3 — complete)
 
 1. **Sandbox fabric** — SandboxBackend protocol + unshare namespace spawner + lifecycle manager (create/list/snapshot/destroy) + concurrency guard + per-sandbox workdir; isolation + resource-limit probes
 2. **Live build telemetry** — TelemetryEvent models + SQLite TraceStore + WebSocket ingest endpoint + seq gap detection + in-sandbox capture agent
 3. **Process-trace grading engine** — deterministic feature/digest computation + rubric scoring via LLM structured output + GradeStore; calibrated against v0.2 mock corpora
 4. **Variant task generation** — template library + seeded LLM instantiation + VariantStore + difficulty normalization anchors
 5. **Oral / voice defense** — VoiceProvider protocol + STT/TTS + mock + browser fallback + Examiner agent + transcript/integrity-signal capture
-6. **Agent re-grounding + learner surface integration** — Lab/Assessor/Proctor consume real telemetry/grades/defense signals; learner sandbox mockup → real in-browser xterm.js build/run; assessment mockup → live defense + live grading
+6. **Agent re-grounding + learner surface integration** — Lab/Assessor/Proctor consume real telemetry/grades/defense signals; learner sandbox mockup → real in-browser build/run (Run/Test buttons executing in a namespace sandbox, read-only exec-output panel — no interactive shell, CUT-2/G-8); assessment mockup → live defense + live grading
 
 ---
 
@@ -184,16 +248,19 @@ The v0.1 build order (monorepo → types → mock data → tokens → primitives
 
 ---
 
-## Future Architecture (Post-v0.3, for reference)
+## Future Architecture (Post-v0.4, for reference)
 
-v0.3 delivers the real credential engines; later milestones fill in the remaining platform:
+v0.4 delivers distribution (CLI + binary releases); later milestones fill in the remaining platform:
 
 - **In-memory sessions → PostgreSQL + Drizzle/SQLModel** — SessionStore + v0.3 TraceStore/GradeStore/VariantStore/DefenseStore protocols swap SQLite→Postgres with no API changes
 - **userns subprocess sandboxes → containerd/runc backend** — D-024 `SandboxBackend` protocol swap; same lifecycle API
-- **Coding-IDE sandbox → design tool + simulation environments** — REQ-F-021 full scope (v0.4)
+- **Coding-IDE sandbox → design tool + simulation environments** — REQ-F-021 full scope (v0.5)
 - **Mock provider → per-agent model routing** — provider factory already selects by config; per-agent `AI_<AGENT>_MODEL` overrides
-- **No auth → real KYC + sessions** — **deferred per founder directive**; REQ-F-017 identity/age-gating lands post-v0.3 (v0.4+). Age-gating remains the v0.1 visual flow mockup
+- **No auth → real KYC + sessions** — **deferred per founder directive; moved to v0.5 with D-016**; REQ-F-017 identity/age-gating lands post-v0.4. Age-gating remains the v0.1 visual flow mockup
+- **Mock voice → real server STT/TTS (openai-audio provider)** — CUT-1/G-7 seam moved to v0.5 per D-016; VoiceProvider protocol is the drop-in point
+- **linux x64 binary → macOS arm64 + auto-update** — D-036 defers non-linux targets (unverifiable on this box); `nextcraft upgrade` (self-replace from latest release) is the natural v0.5+ follow-up
+- **Exec-telemetry seq-lease / replay-margin fix** — the P6-lesson one-line ACK gap moves to v0.5 per D-016
 - **No search → Semantic vector search (pgvector)** — Filter UI replaced with vector similarity search
 - **No payments → Payment processing** — Pricing page replaced with real subscription/payment flows
 
-The monorepo structure (apps/web + apps/ai-service + packages/*) accommodates further apps without restructuring.
+The monorepo structure (apps/web + apps/ai-service + apps/cli + packages/*) accommodates further apps without restructuring.

.ciagent/CHECKPOINT.json

@@ -1,8 +0,0 @@
-{
-  "phase": 1,
-  "stage": "complete",
-  "milestone": "v0.3",
-  "phase_role": "execution",
-  "attempts": 0,
-  "updated_at": "2026-09-11T18:43:22Z"
-}

.ciagent/GRILL.md

@@ -46,3 +46,41 @@ The credential-pipeline architecture (telemetry → trace → grade → defense)
 ## Outcome
 
 **GO** — all six binding decisions and both scope cuts applied to PLAN.md / REQUIREMENTS.md / ROADMAP.md / PROJECT.md before Phase 1 execution. No axis requires escalation (all resolvable at confidence ≥ 0.85). The milestone no longer claims resource enforcement it cannot deliver, and the no-auth abuse vector is closed at MVP scale.
+
+---
+# Nextcraft v0.4 — GRILL.md (Adversarial Review Verdict)
+
+**Stage:** GRILL, Phase 0 pre-execution · **Verdict:** GO-WITH-CHANGES · **Confidence:** 0.83
+
+## Summary
+
+The distribution milestone is small, founder-directed (D-016, confidence 0.99), and additive (zero changes to the running credential pipeline). The plan's central risk: **Node SEA was probe-verified as a flag, not as a working build** — the v0.3 lesson (A-101: probe the mechanism, not the existence) applies. Second gap: a binary whose `--version` lies (stale package.json) would poison the "ongoing binaries" contract. Third: sed-based JSON parsing in install.sh is a fragility + integrity risk. Fourth: "ongoing binaries" has no enforcement mechanism beyond prose. All four closed by binding decisions G-101..G-104 below. No scope cuts required — the milestone is already minimal.
+
+## Per-Axis Findings
+
+| Axis | Verdict | Rationale |
+|------|---------|-----------|
+| Business case | PASS | Founder directive explicit + recorded (D-016). Evidence of need: live Gitea probe shows latest release v0.2.8 with ZERO assets; bootstrap requires repo archaeology (scripts found only via package.json spelunking). |
+| Scope | PASS | 5 REQs, 3 execution phases, one focused surface (apps/cli + scripts). Smallest milestone yet. macOS arm64 already cut (D-036, unverifiable here). |
+| Feasibility | CONCERN (fixed) | SEA flag exists on node v24.15.0, but no end-to-end SEA binary was built during RESEARCH. postject availability assumed (`npx postject` — needs npm registry reachability, unproven). Zipapp fallback requires python3 on target — an honest-degradation ladder, not a silent downgrade. → G-101. |
+| Honest versioning | CONCERN (fixed) | `--version` from package.json would print a stale hardcoded version inside a per-release binary — breaks upgrade detection + the one-liner's re-run-to-upgrade promise. → G-102. |
+| Install integrity | CONCERN (fixed) | sed/grep JSON parsing is brittle; a parse failure must never fall through to installing an unverified artifact. Exact asset-name matching + hard-degrade to source instructions. → G-103. |
+| Sequencing | PASS | P1 CLI (source-runnable) → P2 binary+pipeline → P3 docs+E2E matches dependency order; each phase ships independently. |
+| Cost/quota | PASS | Zero new paid infra; binaries built on-box; Gitea releases free. Dev-only esbuild dep. |
+| Risks | CONCERN (fixed) | Top 3: SEA end-to-end (→ G-101 live probe FIRST in P2), npm registry reachability for esbuild (→ proven by P1's pnpm install must-have), Gitea asset-upload token scope (→ live-proven at the v0.3.2 ship itself). |
+| Adoption/operability | PASS | Consumer = founder + future pilots; one command replaces README archaeology. Rollback trivial (rm ~/.local/bin/nextcraft). No server changes. |
+
+## Binding Decisions (applied to PLAN.md)
+
+- **G-101 (BINDING) — SEA live-build probe is the FIRST P2 action.** Task 2-1-01 builds a real binary before anything depends on it; the build script encodes the fallback ladder explicitly (SEA → zipapp with "requires python3" honesty). If SEA fails on this box, zipapp becomes primary with the docs stating the requirement — no silent claim of node-less operation.
+- **G-102 (BINDING) — Version stamping at build time.** `build-binary` accepts the shipping tag and stamps it into the bundle (`NEXTCRAFT_VERSION` replace); `--version` prints it; install E2E asserts the installed binary reports the tag it was downloaded from. A binary may never report a version it was not built as.
+- **G-103 (BINDING) — Install-script integrity hard-degrade.** install.sh matches assets by EXACT name (`nextcraft-linux-x64`, `nextcraft-linux-x64.sha256`); any parse/lookup/download failure degrades to source-bootstrap instructions (exit 0) — never installs unverified or name-approximate artifacts. Checksum mismatch = hard stop, exit 1, explicit do-not-run message. dash-safe POSIX sh, no jq.
+- **G-104 (BINDING) — Ongoing-binaries enforcement.** Every ship from v0.3.2 onward MUST run `scripts/release-assets.sh <tag>` after tag+merge (best-effort, non-blocking, `release_pending` escalation on failure — but attempted + logged every release). The final-phase audit gate includes "milestone release carries both assets" as a check. This makes the founder's "ongoing binaries" directive a pipeline property, not prose.
+
+## Escalations
+
+None. All four concerns resolved at confidence ≥ 0.85. No axis requires founder escalation (directive already explicit).
+
+## Outcome
+
+**GO** — G-101..G-104 applied to PLAN.md before Phase 1 execution. The milestone claims only what its probes prove, and the ongoing-binaries contract has an enforcement mechanism.

.ciagent/PERSONAS.md

@@ -2,19 +2,19 @@
 
 ## Persona Roster
 
-> **v0.3 update (RESEARCH, lead-developer assessment):** backend-engineer territory extended to the new engine modules (telemetry/grading/variants persistence + APIs). New phase-relevant custom personas added: **sandbox-engineer** (Linux-namespace isolation infra) and **voice-engineer** (STT/TTS + Examiner agent audio pipeline). ai-engineer re-scoped to LLM/agents/prompts + grading/variant/voice *model-facing* logic. **security-auditor stays inactive** (KYC deferred per founder directive). frontend-engineer gains real-sandbox (xterm.js), live-telemetry, and live-defense surfaces.
+> **v0.4 update (RESEARCH, lead-developer assessment):** milestone pivoted to Distribution & Bootstrap CLI (founder directive D-016). New custom persona **cli-engineer** (domain `cli`) owns apps/cli end-to-end: doctor/bootstrap/verify/dev commands, checks, spawn wrappers, the SEA binary build, the one-liner install script, and the release-asset pipeline. backend-engineer retains the scripts/ + turbo/root-package integration surface. **sandbox-engineer and voice-engineer deactivated** (their v0.3 code is complete and untouched this milestone — reason fields below). ai-engineer light-touch (no model-facing work in v0.4). **security-auditor re-activated (phase-specific)** for the install pipeline: curl|bash attack surface, checksum trust, PATH writes, secrets handling in the release flow. frontend-engineer/design-system-engineer/data-engineer inactive (zero UI/data-scope tasks in v0.4 — retained below with reasons).
 
 ### lead-developer
 ```yaml
 active: true
 phase_specific: false
-reason: Coordinates task decomposition across web, AI service, engine, and sandbox territories; resolves conflicts between frontend, backend, AI, sandbox, and voice personas
+reason: Coordinates task decomposition across CLI, scripts, release-pipeline, and docs territories; resolves cli-engineer/backend-engineer boundary (scripts vs CLI)
 domain: coordination
 frameworks:
   - next.js
   - turborepo
   - pnpm
-  - fastapi
+  - node
 constraints:
   - pragmatic
   - battle-tested defaults
@@ -27,99 +27,83 @@ territory:
   - "apps/ai-service/pyproject.toml"
 ```
 
-### frontend-engineer
+### cli-engineer
 ```yaml
 active: true
 phase_specific: false
-reason: Phase 6 real-engine learner-surface integration — xterm.js in-browser terminal, file-tree/run/test controls, live telemetry panels, live voice defense UI, live grading display. Owns all page components, layouts, surface-specific UI.
-domain: frontend
-frameworks:
-  - react
-  - next.js
-  - tailwindcss
-  - lucide-react
-  - recharts
-  - react-flow
-  - "@xterm/xterm"
-  - "@xterm/addon-fit"
-constraints:
-  - component-first
-  - server-components-default
-  - minimal-client-js
-  - sse-client-buffering (buffer bytes, split frames on \n\n, join data: lines)
-  - abortcontroller-cleanup (idempotent abort in effect cleanup)
-  - websocket-lifecycle (typed messages, reconnect backoff, cleanup)
-  - mediarecorder-permission-ux (mic consent, graceful no-mic fallback)
-  - responsive-all-breakpoints
-  - dark-mode-support
-territory:
-  - "apps/web/**"
-  - "packages/ui/**"
-  - "packages/mock-data/**"
-  - "packages/types/**"
-```
-
-### data-engineer
-```yaml
-active: true
-phase_specific: false
-reason: Owns TS mock data layer schema and typed definitions + TS types for telemetry/trace/grade/variant/defense shapes the web surfaces consume. Does NOT own the Python corpus or engine stores — aligned by convention (D-021).
-domain: data
+reason: v0.4 custom persona (RESEARCH) — owns the distribution milestone core: nextcraft CLI (doctor/bootstrap/verify/dev), pure check logic, spawn wrappers with timeouts, Node SEA binary build (D-033), one-liner install.sh (D-035), checksum sidecar, and Gitea release-asset upload (D-036)
+domain: cli
 frameworks:
+  - node
   - typescript
+  - node:test
+  - esbuild
+  - node-sea
+  - posix-sh
 constraints:
-  - schema-first
-  - type-safe
-  - migration-ready
-  - mock-data-only
+  - stdlib-only-runtime (no runtime npm deps; esbuild dev-only)
+  - thin-wrapper (never re-implement scripts/bootstrap.sh or dev.sh — compose via spawn, A-202/A-209)
+  - timeout-every-spawn (no unbounded subprocess)
+  - actionable-errors (every failed check tells the user how to fix it)
+  - graceful-degradation (install never hard-fails; source-bootstrap fallback, A-206)
+  - checksum-before-install (sha256 verify before chmod+install, A-207)
+  - secrets-never-in-cli (no key generation; .env.example -> .env copy only, A-210)
+  - fail-loud-exit-codes (0 ok / 1 failure / 2 usage)
 territory:
-  - "packages/types/**"
-  - "packages/mock-data/**"
+  - "apps/cli/**"
+  - "scripts/install.sh"
+  - "scripts/release-assets.sh"
 ```
 
 ### backend-engineer
 ```yaml
 active: true
 phase_specific: false
-reason: Owns apps/ai-service app shell, config, API endpoints (incl. WebSocket telemetry ingest), engine persistence (SQLite stores), scripts, and test harness. Extended for v0.3 engine modules.
+reason: Owns the script + monorepo integration surface the CLI composes: apps/ai-service/scripts/*, root package.json cli:* passthrough scripts, turbo task wiring (D-037/D-022). Python ai-service itself is untouched this milestone (v0.3 complete).
 domain: backend
 frameworks:
   - fastapi
-  - uvicorn
-  - pydantic
-  - pydantic-settings
-  - httpx
-  - pytest
-  - sqlmodel
-  - sqlalchemy
-  - websockets
-  - aiofiles
+  - bash
+  - turborepo
+  - pnpm
 constraints:
-  - provider-agnostic-boundaries (engine modules import nothing from agents/ or api/)
-  - streaming-first
-  - sqlite-first-persistence (protocol-wrapped stores, Postgres-ready, D-027)
-  - secrets-via-env-only
-  - mock-provider-in-tests
-  - websocket-contract (typed envelopes, seq gap detection, D-026)
+  - scripts-are-truth (bootstrap.sh/dev.sh stay the single source of bootstrap orchestration; CLI only wraps)
+  - idempotent-scripts (re-runnable without side effects)
+  - secrets-via-env-only (D-014; dev.sh exports from .ciagent/.env.secrets)
 territory:
-  - "apps/ai-service/ai_service/main.py"
-  - "apps/ai-service/ai_service/config.py"
-  - "apps/ai-service/ai_service/api/**"
-  - "apps/ai-service/ai_service/telemetry/store.py"
-  - "apps/ai-service/ai_service/telemetry/ingest.py"
-  - "apps/ai-service/ai_service/grading/store.py"
-  - "apps/ai-service/ai_service/variants/store.py"
   - "apps/ai-service/scripts/**"
   - "apps/ai-service/package.json"
-  - "apps/ai-service/tests/api/**"
+  - "package.json"
   - "turbo.json"
+  - ".gitignore"
+```
+
+### security-auditor
+```yaml
+active: true
+phase_specific: true
+reason: v0.4 re-activated (phase-specific) — the install pipeline is the first externally-consumed attack surface: curl|bash piping, latest-release resolution, checksum trust root, PATH writes to ~/.local/bin, download tempdir hygiene, release-asset upload token handling. No KYC/PII work (still v0.5).
+domain: security
+frameworks:
+  - posix-sh
+  - curl
+  - sha256sum
+constraints:
+  - STRIDE-classified
+  - no-pipe-to-shell-without-checksum (download -> verify -> install order)
+  - tmpdir-safe (mktemp, no predictable paths, trap cleanup)
+  - token-never-echoed (release upload resolves .env* only, never logs)
+territory:
+  - "scripts/install.sh"
+  - "scripts/release-assets.sh"
+  - "apps/cli/src/lib/spawn.ts"
 ```
 
 ### ai-engineer
 ```yaml
 active: true
 phase_specific: false
-reason: Owns the LLM provider layer, agent framework, prompt library, structured outputs, and the model-facing logic of v0.3 engines — trace-digest→rubric grading prompts (grading/features.py+engine.py), variant instantiation (variants/templates.py+generator.py), and the Examiner agent. Owns the deterministic-mock corpora.
+reason: Light-touch v0.4 — no model-facing work in the distribution milestone; retained to guard the CLI against touching agent/engine boundaries and to keep territory mappings accurate for v0.5 (voice real-path, seq-lease).
 domain: ai
 frameworks:
   - pydantic
@@ -127,116 +111,101 @@ frameworks:
   - pytest
 constraints:
   - provider-agnostic-protocol
-  - prompts-are-code
-  - json-defensive-parsing
   - never-call-cloud-in-tests
-  - delta-passthrough
-  - llm-sees-digest-not-raw-trace (D-028)
-  - seeded-variant-reproducibility (D-029)
 territory:
   - "apps/ai-service/ai_service/llm/**"
   - "apps/ai-service/ai_service/agents/**"
   - "apps/ai-service/ai_service/prompts/**"
-  - "apps/ai-service/ai_service/corpus/**"
-  - "apps/ai-service/ai_service/grading/features.py"
-  - "apps/ai-service/ai_service/grading/engine.py"
-  - "apps/ai-service/ai_service/variants/templates.py"
-  - "apps/ai-service/ai_service/variants/generator.py"
-  - "apps/ai-service/tests/llm/**"
-  - "apps/ai-service/tests/agents/**"
 ```
 
-### sandbox-engineer
+### frontend-engineer
 ```yaml
-active: true
-phase_specific: true
-reason: v0.3 custom persona (RESEARCH) — owns the sandbox fabric: SandboxBackend protocol, unshare-based Linux user/mount/pid/net namespace spawner, per-sandbox workdir, resource limits, lifecycle manager, concurrency guard, and the in-sandbox capture agent. Probe-verified isolation on this box (D-024).
-domain: infra
+active: false
+phase_specific: false
+reason: v0.4 has zero UI-scope work (no web/pages/components changes planned in the distribution milestone); v0.3 surfaces are complete. Reactivated at v0.5 when deferred UX work resumes.
+domain: frontend
 frameworks:
-  - python
-  - linux-namespaces
-  - asyncio
-  - pytest
+  - react
+  - next.js
+  - tailwindcss
 constraints:
-  - isolation-verified (probe must show in-ns uid=0, network isolated, writes to workdir only)
-  - backend-protocol-swap (no containerd assumption; D-024)
-  - resource-limits-enforced (cpu/mem/time quotas observable)
-  - no-daemon (subprocess-only; no docker/containerd service)
-  - capacity-guard (1-5 concurrent; 503 when full, D-032)
+  - component-first
+  - server-components-default
 territory:
-  - "apps/ai-service/ai_service/sandbox/**"
-  - "apps/ai-service/scripts/sandbox-agent.py"
-  - "apps/ai-service/tests/sandbox/**"
-```
-
-### voice-engineer
-```yaml
-active: true
-phase_specific: true
-reason: v0.3 custom persona (RESEARCH) — owns the voice layer: VoiceProvider protocol, STT/TTS against a compatible endpoint, deterministic mock (tests never call a voice API), browser-native fallback, and the media-path wiring consumed by the Examiner agent and assessment UI.
-domain: ai-media
-frameworks:
-  - pydantic
-  - httpx
-  - pytest
-  - web-mediarecorder
-constraints:
-  - provider-agnostic-protocol (D-030)
-  - never-call-voice-api-in-tests
-  - browser-native-fallback (no-key path still functions)
-  - bounded-turn-latency (conversational feel budget)
-territory:
-  - "apps/ai-service/ai_service/voice/**"
-  - "apps/ai-service/tests/voice/**"
+  - "apps/web/**"
+  - "packages/ui/**"
 ```
 
 ### design-system-engineer
 ```yaml
-active: true
+active: false
 phase_specific: false
-reason: Owns the shared component library, design tokens, and visual consistency. v0.3 duty: new primitives for the real build/assessment surfaces (terminal frame, telemetry status indicator, mic/record control, grade badge, defense transcript viewer).
+reason: No design-token or primitive work in v0.4; roster retained for v0.5.
 domain: frontend
 frameworks:
   - tailwindcss
   - storybook
-  - lucide-react
 constraints:
   - design-token-driven
   - wcag-aa-contrast
-  - dark-mode-required
-  - consistent-across-surfaces
 territory:
   - "packages/ui/**"
 ```
 
-### security-auditor
+### data-engineer
 ```yaml
 active: false
 phase_specific: false
-reason: Identity/age-gating (KYC) deferred beyond v0.3 per founder directive (A-110) — no real auth or PII backend lands this milestone. Security coverage remains: verifier's STRIDE layer + Phase 7 secrets-hygiene checklist (keys absent from code/logs/commits/errors, localhost-only CORS, no PII in prompts). Sandbox isolation safety is owned by sandbox-engineer's probe-verified constraint.
-domain: security
+reason: No schema/mock-data work in v0.4; types packages untouched. Reactivated if CLI surfaces need shared types (not planned — CLI is self-contained).
+domain: data
+frameworks:
+  - typescript
+constraints:
+  - schema-first
+  - type-safe
+territory:
+  - "packages/types/**"
+  - "packages/mock-data/**"
+```
+
+### sandbox-engineer
+```yaml
+active: false
+phase_specific: false
+reason: v0.3 persona — sandbox fabric shipped complete (v0.2.x series); v0.4 touches no sandbox code. doctor only *checks* unshare availability; no sandbox logic changes. Reactivated at v0.5 (design/sim environments).
+domain: infra
+frameworks:
+  - python
+  - linux-namespaces
+constraints: []
+territory:
+  - "apps/ai-service/ai_service/sandbox/**"
+```
+
+### voice-engineer
+```yaml
+active: false
+phase_specific: false
+reason: v0.3 persona — voice defense shipped complete (mock-first, CUT-1); real server STT/TTS moved to v0.5 per D-016. No v0.4 voice work.
+domain: ai-media
 frameworks: []
 constraints: []
-territory: []
+territory:
+  - "apps/ai-service/ai_service/voice/**"
 ```
 
 ## Phase-Specific Personas
 
 | Persona | Phases | Removed After |
 |---------|--------|---------------|
-| sandbox-engineer | 1 (primary), 2, 6 | persists while sandbox fabric exists |
-| voice-engineer | 5 (primary), 6 | persists while voice defense exists |
+| security-auditor | 2 (primary: install pipeline), 3, 4 (final review) | milestone complete |
 
-All other active personas span the entire milestone. data-engineer and design-system-engineer are light-touch outside their phases.
+All other personas span the milestone. Deactivated personas receive no tasks.
 
 ## Territory Conflict Resolution
 
 | Conflict | Resolution |
 |----------|------------|
-| frontend-engineer vs data-engineer (packages/types, packages/mock-data) | data-engineer owns type definitions and mock data schema (incl. new telemetry/grade/variant/defense TS types); frontend-engineer consumes them. |
-| frontend-engineer vs design-system-engineer (packages/ui) | design-system-engineer owns design tokens and primitive components (terminal frame, mic control, grade badge); frontend-engineer owns composite components and page-level UI. |
-| ai-engineer vs backend-engineer (grading/variants) | ai-engineer owns the model-facing files (features/engine/templates/generator = LLM logic + prompts); backend-engineer owns the persistence stores + API endpoints. Boundary: stores are pure SQLite; engine logic is pure compute. |
-| sandbox-engineer vs backend-engineer (sandbox/) | sandbox-engineer owns `ai_service/sandbox/**` + capture agent; backend-engineer owns the API route that composes `sandbox/manager.py` via DI. manager.py has a narrow typed interface consumed by api/. |
-| voice-engineer vs ai-engineer (Examiner agent) | ai-engineer owns `agents/examiner.py` + its prompt; voice-engineer owns `voice/**` (audio in/out). Examiner calls `voice/` through the `VoiceProvider` protocol — never imports concrete providers. |
-| ai-engineer vs data-engineer (mock duplication) | ai-engineer owns `ai_service/corpus/` (Python); data-engineer owns `packages/mock-data` (TS). Shared IDs/shapes aligned by convention (D-021). |
-| lead-developer vs any | lead-developer coordinates only, does not directly modify code files. |
+| cli-engineer vs backend-engineer (scripts/) | backend-engineer owns `apps/ai-service/scripts/**` + root `package.json`/`turbo.json` wiring; cli-engineer owns `apps/cli/**` + top-level `scripts/install.sh` + `scripts/release-assets.sh` and *consumes* backend scripts via spawn — never edits them |
+| cli-engineer vs security-auditor (install.sh) | cli-engineer implements; security-auditor reviews + may patch security defects directly in install.sh/spawn.ts (its territory) |
+| lead-developer vs any | lead-developer coordinates only, does not directly modify code files |

.ciagent/PLAN.md

@@ -1,438 +1,210 @@
-# Nextcraft v0.3 — PLAN.md
+# Nextcraft v0.4 — PLAN.md
 
 ## Overview
 
-This plan covers execution phases 1-6 of milestone v0.3 (Credential Engines): the real engines that replace v0.2's mock inputs — a namespace-isolated sandbox fabric, live build telemetry over WebSocket + SQLite, a process-trace grading engine, seeded per-learner variant task generation, and an oral/voice defense with a seventh Examiner agent — plus re-grounding the Lab/Assessor/Proctor agents onto real engine inputs and wiring the v0.1 learner surfaces to the real build/defense/grading paths. Phases are strictly sequential (P1→P6); within each phase, Wave 1 tasks are parallelizable (no cross-file dependencies) and later waves depend on earlier ones.
+This plan covers execution phases 1–3 of milestone v0.4 (Distribution & Bootstrap CLI) plus the final phase (P4 review+ship). The milestone delivers the founder directive (D-016): a streamlined install for Nextcraft — a `nextcraft` bootstrap CLI shipped as a linux x64 binary, installed via a one-liner script, with binaries published on **every ongoing release** from v0.4 onward. Phases are strictly sequential (P1→P3); within each phase, Wave 1 tasks are parallelizable (no cross-file dependencies) and later waves depend on earlier ones.
 
-**Environment facts (apply throughout):** Python 3.11.2 via `python3 -m venv` (no uv, no system pip); pnpm 12.3.4 via corepack; turborepo; ai-service port **8420**; default model `gemma4:31b` (config via `AI_TUTOR_MODEL`); ollama-cloud base `https://ollama.com/v1` (OpenAI-compatible, Bearer auth); keys live only in gitignored `.ciagent/.env.secrets` (exported by `scripts/dev.sh`) — never in code, commits, or logs; all automated tests use the deterministic mock LLM and mock voice providers and **never call cloud or voice APIs**. New ai-service deps this milestone: `sqlmodel`, `sqlalchemy`, `websockets`, `aiofiles` (all PyPI-verified; added in Task 1-1-02). SQLite data (`apps/ai-service/ai_service/data/`) and sandbox dirs (`apps/ai-service/sandboxes/`) are already gitignored. **KYC/age-gating is deferred per founder directive (A-110)** — no identity work; age-gating stays the v0.1 visual flow mockup. **GRILL scope decisions (this revision):** real server STT/TTS (`OpenAIAudioProvider`) deferred to v0.4 — voice defense is mock+browser-first (CUT-1/G-7); the interactive xterm.js shell relay is deferred to v0.4 — the build panel is run/test-buttons + read-only exec output (CUT-2/G-8), so `@xterm/*` is NOT a v0.3 dependency; sandbox abuse control (per-learner caps + allowlist, G-5) ships even though KYC is deferred; sandbox resource limits are partially enforced (memory/CPU/wall-clock + workdir-size sweep; per-sandbox pids and hard disk quota are accepted gaps, G-1/G-2).
+**Environment facts (probe-verified, apply throughout):** Go MISSING, Rust MISSING, gcc 12.2 present, **node v24.15.0 x64 linux (SEA-capable)**, python3 3.11.2, `shasum` 6.02, pnpm 12.3.4 via corepack, turborepo 2.3.3, tsx 4.23 in root devDeps path. Gitea API verified live at `https://git.coreci.dev/api/v1` (latest release v0.2.8, **zero assets** — the gap this milestone closes). Existing orchestration: `apps/ai-service/scripts/bootstrap.sh` (idempotent venv+pip incl. the no-ensurepip get-pip path), `apps/ai-service/scripts/dev.sh` (secrets export → uvicorn :8420), `apps/ai-service/.env.example` (full AI_* template). Root scripts: `ai:dev/ai:test/ai:bootstrap/ai:lint` turbo passthroughs (D-022 pattern to mirror as `cli:*`). Secrets live only in gitignored `.ciagent/.env.secrets` (GITEA_TOKEN, OLLAMA_API_KEY, OLLAMA_BASE_URL) — never in code, commits, or logs; tests never call the cloud or the forge (mocks/fixtures only).
+
+**Milestone type:** feature. Tags: phase 0 → **v0.3.0**, P1 → v0.3.1, P2 → v0.3.2, P3 → v0.3.3, final phase P4 → **v0.3.4 = milestone release**. **GRILL binding decisions (this revision):** G-101 — SEA live-build probe is the FIRST P2 action (mechanism, not flag, must be proven); fallback ladder encoded honestly (zipapp requires python3 on target). G-102 — binary `--version` stamped from the shipping tag at build time (never a stale package.json version); install E2E asserts the installed binary reports its release tag. G-103 — install.sh matches assets by exact name; any parse/download failure degrades to source-bootstrap instructions (exit 0), never installs unverified artifacts; checksum mismatch = hard stop exit 1. G-104 — every ship from v0.3.2 onward runs `scripts/release-assets.sh <tag>` (best-effort, logged, non-blocking); the P4 audit gate checks the milestone release carries both assets.
 
 | Phase | Name | Requirements | Waves | Personas |
 |-------|------|-------------|-------|----------|
-| 1 | Sandbox fabric | REQ-3-001, 002 | 3 | sandbox-engineer, backend-engineer, ai-engineer (W1 lint only) |
-| 2 | Live build telemetry | REQ-3-003 | 4 | sandbox-engineer, backend-engineer, data-engineer |
-| 3 | Process-trace grading engine | REQ-3-004 | 3 | ai-engineer, backend-engineer |
-| 4 | Variant task generation | REQ-3-005 | 3 | ai-engineer, backend-engineer, data-engineer |
-| 5 | Oral / voice defense | REQ-3-006 | 4 | voice-engineer, ai-engineer, backend-engineer |
-| 6 | Agent re-grounding + learner surface integration | REQ-3-007, 008 | 5 | ai-engineer, frontend-engineer, design-system-engineer, data-engineer, backend-engineer, lead-developer |
-
----
-
-## Phase 1: Sandbox Fabric
-
-**Requirements:** REQ-3-001, REQ-3-002
-**Goal:** `SandboxBackend` protocol + `unshare`-based namespace spawner (D-024) + lifecycle manager with concurrency guard (D-032) + per-sandbox workdir; isolation and resources probe-verified on this box; `/v1/sandboxes` API live; deps + gitignore landed
-
-### Wave 1: Foundations (parallel — no shared files)
-
-#### Task 1-1-01: SandboxBackend protocol + unshare spawner + probe test
-- **Persona:** sandbox-engineer — **REQ:** REQ-3-001, REQ-3-002
-- **Files:** `apps/ai-service/ai_service/sandbox/__init__.py`, `apps/ai-service/ai_service/sandbox/backend.py`, `apps/ai-service/ai_service/sandbox/workdir.py`, `apps/ai-service/ai_service/sandbox/unshare_backend.py`, `apps/ai-service/tests/sandbox/__init__.py`, `apps/ai-service/tests/sandbox/test_isolation.py`
-- **Action:** `backend.py`: `SandboxBackend` protocol + `SandboxSpec` (sandbox_id, learner_id, workdir, resource limits) + `SandboxHandle` (id, pid, workdir, created_at); `spawn(spec)`, `exec(handle, cmd)`, `snapshot(handle) -> Path`, `destroy(handle)`. `workdir.py`: per-sandbox layout under `apps/ai-service/sandboxes/<id>/` (workspace/ writable, snapshot() = recursive copy to `snapshots/<ts>/`) — no symlinks as the snapshot mechanism. `unshare_backend.py`: subprocess spawner — `unshare --user --map-root-user --mount --pid --fork --net` with the per-sandbox dir bind-mounted (`--bind <dir> /work`) and `chdir /work` (D-024); pipes for stdout/stderr; async wrappers. `test_isolation.py` — **re-verify box isolation properties (runs on this box, guarded by probe skip):** (a) `id -u` inside namespace prints `0`; (b) `ip link` inside namespace shows 0 usable interfaces (loopback-only/no carrier) — network isolated; (c) file written to `/work/inside.txt` lands at `sandboxes/<id>/workspace/inside.txt` on the host; (d) attempt to write outside the mount (e.g. host tmp path via bind) does not escape the per-sandbox dir; (e) `/proc` visibility degraded (proc-remount not permitted per A-101 — assert the probe documents this, not that it fails).
-- **Verify:** `pnpm ai:test` — `tests/sandbox/test_isolation.py` green on this box (probe-gated: skips with an explicit reason if userns unavailable); `lint` clean
-
-#### Task 1-1-02: v0.3 dependencies + gitignore + config additions
-- **Persona:** backend-engineer — **REQ:** REQ-3-001
-- **Files:** `apps/ai-service/pyproject.toml` (update), `apps/ai-service/ai_service/config.py` (update), `apps/ai-service/.env.example` (update), `apps/ai-service/scripts/bootstrap.sh` (update if needed), root `package.json` (no change), `turbo.json` (no change)
-- **Action:** Add pinned deps: `sqlmodel`, `sqlalchemy`, `websockets`, `aiofiles` to pyproject. `config.py` additions (env_prefix `AI_`): `AI_SANDBOX_DIR` (default `apps/ai-service/sandboxes`), `AI_SANDBOX_MAX_CONCURRENT` (default 5, D-032), `AI_SANDBOX_CPU_LIMIT` (default 1 core / cpu.max), `AI_SANDBOX_MEM_LIMIT_MB` (default 512), `AI_SANDBOX_PIDS_LIMIT` (default 256), `AI_SANDBOX_TIMEOUT_S` (default 1800), `AI_DB_PATH` (default `ai_service/data/nextcraft.db`), `AI_VOICE_BASE_URL` / `AI_VOICE_API_KEY` / `AI_VOICE_STT_MODEL` / `AI_VOICE_TTS_MODEL` (all **optional**, default empty — mock-first, D-030; documented in `.env.example` and README). Confirm `.gitignore` already covers `ai_service/data/` + `sandboxes/` (it does — v0.3 block present). Re-run bootstrap idempotently to install new deps.
-- **Verify:** `pnpm ai:bootstrap` re-installs cleanly (no-op venv, new wheels land); `python -c "import sqlmodel, sqlalchemy, websockets, aiofiles"` succeeds in the venv; settings parse with new keys unset
-
-#### Task 1-1-03: Resource-limit probe documentation + harness ruff pass
-- **Persona:** ai-engineer — **REQ:** REQ-3-001
-- **Files:** `apps/ai-service/README.md` (update: sandbox section + probe transcript), `apps/ai-service/pyproject.toml` (no change), `apps/ai-service/tests/sandbox/test_isolation.py` (no change)
-- **Action:** Record the A-101 probe transcript in README (verbatim commands + observed output from this box: `unshare --user --map-root-user --mount --pid --fork --net id -u` → `0`; `ip link` → loopback only; write containment). Document the v0.3 resource-limit mechanism choice: cgroup-v2 delegation via per-sandbox scope files is **not available** on this box without sudo → enforcement = subprocess-level (`ulimit`-equivalent via `preexec_fn`: RLIMIT_AS for memory, RLIMIT_CPU for CPU-seconds, RLIMIT_NPROC for pids) + hard wall-clock timeout kill in the manager. This is the locked v0.3 mechanism (D-024 + no-sudo constraint). Run ruff over the new tree; fix all findings.
-- **Verify:** `pnpm ai:lint` exits 0; README shows the probe transcript and the rlimit mechanism note
-
-### Wave 2: Lifecycle manager (depends on Wave 1)
-
-#### Task 1-2-01: Sandbox manager + concurrency guard + snapshots
-- **Persona:** sandbox-engineer — **REQ:** REQ-3-001, REQ-3-002
-- **Files:** `apps/ai-service/ai_service/sandbox/manager.py`, `apps/ai-service/tests/sandbox/test_manager.py`
-- **Action:** `SandboxManager`: `create(learner_id) -> SandboxHandle` (guard: active count ≥ `AI_SANDBOX_MAX_CONCURRENT` → raise `PoolFullError` → API maps to **503**, D-032; no queue); `list() -> list[SandboxHandle]`; `get(id)`; `snapshot(id) -> Path` (delegates to workdir); `destroy(id)` (kill process tree, keep or purge workdir per flag); `reap_expired()` background hook for `AI_SANDBOX_TIMEOUT_S` which also performs a **workdir-size sweep**: any sandbox whose `workdir` exceeds `AI_SANDBOX_MAX_WORKDIR_MB` (new config, default 512) is snapshotted-then-destroyed and the event logged as an integrity signal (G-2 — soft disk cap, best-effort, not kernel-enforced); the sweep runs on the same timer as the timeout reaper. Enforce rlimits per spawner (Task 1-1-03: RLIMIT_AS + RLIMIT_CPU + RLIMIT_FSIZE=50MB as a cheap single-file disk guard (a-2); RLIMIT_NPROC noted as shared-per-host-uid, not relied on (G-1)) at exec time. Handle registry persisted **in-memory** (v0.3, single process; not a store — see D-019 precedent) with a clear note that handles are process-local. Startup reaper (a-1): on lifespan boot, scan `AI_SANDBOX_DIR` for workdirs whose recorded pid is dead and reap them, logging a warning. Narrow typed interface only — manager never imports api/ (boundary rule).
-- **Verify:** `pnpm ai:test` — test_manager covers create/list/destroy/snapshot, 6th create raises PoolFullError (503 path), destroy kills the namespace process (pid gone), snapshot dir exists with workspace contents, timeout reaper removes a stale handle
-
-#### Task 1-2-02: Resource-limit enforcement test
-- **Persona:** sandbox-engineer — **REQ:** REQ-3-002
-- **Files:** `apps/ai-service/tests/sandbox/test_resource_limits.py`
-- **Action:** Concrete enforcement probes (guarded like isolation tests): (a) spawn a process that allocates > `RLIMIT_AS` → assert it dies with MemoryError/killed within a bound; (b) spawn a CPU spinner past `RLIMIT_CPU` → assert SIGXCPU/kill; (c) single huge file > `RLIMIT_FSIZE` → assert write failure (a-2 partial disk guard); (d) wall-clock: spawn `sleep 9999` with a small manager timeout → reaper destroys it; (e) **disk sweep (G-2)**: write > `AI_SANDBOX_MAX_WORKDIR_MB` across many files → assert the manager sweep destroys the sandbox and logs the integrity signal. Assert limits are observable (handle reports its limit set). NOTE (G-1): per-sandbox `RLIMIT_NPROC` is shared at the host uid — the fork-bomb probe is documented as shared-budget behavior, NOT asserted as per-sandbox isolation.
-- **Verify:** `pnpm ai:test` — test_resource_limits green; limits proven enforced and observable
-
-### Wave 3: API exposure (depends on Wave 2)
-
-#### Task 1-3-01: Sandboxes API module
-- **Persona:** backend-engineer — **REQ:** REQ-3-001, REQ-3-002
-- **Files:** `apps/ai-service/ai_service/api/sandboxes.py`, `apps/ai-service/ai_service/main.py` (update: include router + lifespan manager), `apps/ai-service/ai_service/api/deps.py` (update), `apps/ai-service/tests/api/test_sandboxes.py`
-- **Action:** DI exposes a singleton `SandboxManager`. Endpoints: `POST /v1/sandboxes {learner_id}` → 201 handle (503 when pool full); `GET /v1/sandboxes` → list; `GET /v1/sandboxes/{id}` → handle; `POST /v1/sandboxes/{id}/snapshot` → snapshot path; `DELETE /v1/sandboxes/{id}` → 204. All behind localhost CORS (A-008). Lifespan creates/destroys the manager; on shutdown destroys any live sandboxes (no orphans). **Abuse control (G-5, NOT KYC):** even in no-auth v0.3 the sandbox API enforces per-`learner_id` rate limiting (`AI_SANDBOX_MAX_PER_LEARNER`, default 1 active → 429) and a global create-rate cap (`AI_SANDBOX_CREATES_PER_MIN`, default 10 → 429); `learner_id` is validated against a server-side allowlist from config (`AI_LEARNER_ALLOWLIST`, default the single mock pilot id → unknown ids rejected 403). This ships in the no-auth milestone so a rogue local process can't exhaust shared NPROC/disk.
-- **Verify:** `pnpm ai:test` — test_sandboxes green (create→list→snapshot→delete roundtrip via TestClient; 6th create → 503; delete of unknown id → 404; abuse control (G-5): non-allowlisted learner_id → 403; >1 active sandbox for one learner → 429; burst of >10 creates/min → 429); manual probe: `curl -X POST localhost:8420/v1/sandboxes -d '{"learner_id":"l1"}'` returns a handle id
-
-### Must-Haves (Phase 1)
-- [ ] Isolation probe test green on this box: in-namespace uid=0, network isolated (0 usable interfaces), host writes confined to the per-sandbox bind dir (A-101 re-verified as an automated test, not just research notes)
-- [ ] Resource limits enforced + observable: memory (RLIMIT_AS) + CPU (RLIMIT_CPU) rlimits kill violating processes; wall-clock reaper destroys stale sandboxes; disk usage capped by a periodic workdir-size sweep in the manager (soft cap, configurable `AI_SANDBOX_MAX_WORKDIR_MB`, default 512MB — NOT kernel-enforced); per-sandbox NPROC is shared across sandboxes at the host uid — documented, not relied on for isolation (G-1, G-2 — test_resource_limits green)
-- [ ] No cross-tenant access: sandbox A cannot read sandbox B's workdir (isolation test asserts containment)
-- [ ] Lifecycle API works end-to-end: create/list/snapshot/destroy via TestClient; pool full → **503** (D-032, no queue)
-- [ ] Snapshot produces a restorable directory copy under the sandbox's own snapshots/ dir
-- [ ] `pnpm ai:test` and `pnpm ai:lint` green; new deps (`sqlmodel`, `sqlalchemy`, `websockets`, `aiofiles`) installed via idempotent bootstrap
-- [ ] Boundary rules hold: `sandbox/` imports nothing from `api/` or `agents/`; only `api/sandboxes.py` composes the manager via DI
-- [ ] No docker/podman/sudo anywhere in the spawner path (D-024); `SandboxBackend` protocol is the only coupling to the spawner (containerd swap possible later)
-
----
-
-## Phase 2: Live Build Telemetry
-
-**Requirements:** REQ-3-003
-**Goal:** First real persistence (D-027: SQLite via SQLModel) with a `TraceStore` protocol; `TelemetryEvent` model with per-(learner,task) monotonic `seq`; WebSocket ingest endpoint (D-026) with gap detection; stdlib-only in-sandbox capture agent streams real sandbox activity into ai-service; trace retrievable by learner+task
-
-### Wave 1: Models + stores + TS types (parallel — no shared files)
-
-#### Task 2-1-01: Telemetry event models
-- **Persona:** backend-engineer — **REQ:** REQ-3-003
-- **Files:** `apps/ai-service/ai_service/telemetry/__init__.py`, `apps/ai-service/ai_service/telemetry/models.py`, `apps/ai-service/tests/telemetry/__init__.py`, `apps/ai-service/tests/telemetry/test_models.py`
-- **Action:** Pydantic/SQLModel `TelemetryEvent`: `learner_id`, `task_id`, `seq` (int, monotonic per (learner,task)), `kind` (`command` | `file_diff` | `run_result` | `test_result` | `activity` | `stdin` | `stdout`), `payload` (JSON), `ts` (datetime, monotonic-envelope), `sandbox_id`. `TraceSpan` derived view (ordered events for one (learner,task)). Validation: seq ≥ 0, kind enum, non-empty ids.
-- **Verify:** `pnpm ai:test` — test_models green (validation rules enforced, JSON payload roundtrip)
-
-#### Task 2-1-02: TraceStore protocol + SQLite implementation
-- **Persona:** backend-engineer — **REQ:** REQ-3-003
-- **Files:** `apps/ai-service/ai_service/telemetry/store.py`, `apps/ai-service/tests/telemetry/test_store.py`, `apps/ai-service/ai_service/data/.gitkeep`
-- **Action:** `TraceStore` protocol (D-027, Postgres-migration-ready): `append(event) -> None` (idempotent on (learner,task,seq) — at-least-once dedup), `get_trace(learner_id, task_id) -> list[TelemetryEvent]` (ordered by seq), `gaps(learner_id, task_id) -> list[int]` (missing seqs), `latest_seq(learner_id, task_id) -> int`, `list_tasks(learner_id) -> list[str]`, `close()`. `SQLiteTraceStore(SQLModel)`: single `telemetry_event` table, composite PK ((learner_id, task_id, seq)), indexes on (learner_id, task_id). Engine creation from `AI_DB_PATH`; `SQLModel.metadata.create_all` at app lifespan. Enable `PRAGMA journal_mode=WAL` + `synchronous=NORMAL` at engine creation (a-3) so concurrent ingest (writer) and trace reads (grader) don't hit `database is locked` under concurrent sandboxes.
-- **Verify:** `pnpm ai:test` — test_store green (append/ordered-get/dedup-on-retry/gap detection/latest_seq; tmp-path SQLite per test)
-
-#### Task 2-1-03: TS types for telemetry/traces
-- **Persona:** data-engineer — **REQ:** REQ-3-003
-- **Files:** `packages/types/telemetry.ts` (new), `packages/types/index.ts` (update)
-- **Action:** TS `TelemetryEvent`, `TraceSpan`, `TelemetryKind` mirroring the Python model field-for-field (cross-referencing header, same string enums). Consumed by Phase 6 web surfaces; no runtime code.
-- **Verify:** `pnpm typecheck` passes; TS type keys match Python model keys exactly
-
-### Wave 2: Capture agent + ingest (depends on Wave 1)
-
-#### Task 2-2-01: In-sandbox capture agent (stdlib-only)
-- **Persona:** sandbox-engineer — **REQ:** REQ-3-003
-- **Files:** `apps/ai-service/scripts/sandbox-agent.py`, `apps/ai-service/tests/sandbox/test_sandbox_agent.py`
-- **Action:** Tiny standalone process (D-031, **stdlib only** — no deps shipped into the namespace): wraps a shell inside the sandbox; captures commands, file diffs (mtime/content polling of `workspace/` at 250ms), run/test results, activity; assigns per-(learner,task) `seq`; buffers to a local spool file on disconnect (at-least-once, D-026); reconnects with **exponential backoff** and flushes spool in order; small WebSocket client implemented over raw `socket` (RFC6455 client handshake + frames — stdlib only, no `websockets` in-namespace). Configured via env baked at spawn (`NC_LEARNER_ID`, `NC_TASK_ID`, `NC_INGEST_URL`).
-- **Verify:** `pnpm ai:test` — unit tests with a loopback fake WS server: ordered seq emission, spool-on-disconnect, reconnect flush preserves order (no loss, dupes deduped server-side), no third-party imports in the file (asserted by AST scan)
-
-#### Task 2-2-02: WebSocket ingest endpoint
-- **Persona:** backend-engineer — **REQ:** REQ-3-003
-- **Files:** `apps/ai-service/ai_service/telemetry/ingest.py`, `apps/ai-service/ai_service/api/sandboxes.py` (update: register WS route), `apps/ai-service/ai_service/main.py` (update), `apps/ai-service/tests/api/test_telemetry_ingest.py`
-- **Action:** `WS /v1/telemetry/ingest` (D-026): accepts connections carrying learner_id/task_id/sandbox_id; validates + appends events to `TraceStore` (idempotent — server-side dedup on (learner,task,seq)); emits **gap warnings** when seq skips (logged + surfaced in a per-connection status); ping/pong keepalive. **Backpressure / flood control (G-3 — replaces silent drop):** bounded inbound queue; on overflow OR when a per-connection cap `AI_TELEMETRY_MAX_EVENTS_PER_TASK` (default 50000) is exceeded → **reject with a 1008 policy-violation close and mark the (learner,task) trace `INCOMPLETE_FLOODED`** (an integrity signal consumed by Proctor). Silent drop-oldest is FORBIDDEN because it corrupts grading input and is indistinguishable from trace-gaming. `GET /v1/telemetry/traces/{learner_id}/{task_id}` returns the ordered trace; `GET /v1/telemetry/gaps/{learner_id}/{task_id}` returns missing seqs.
-- **Verify:** `pnpm ai:test` — test_telemetry_ingest green (TestClient websocket: connect → send 3 events → trace retrievable ordered; resend event 2 → deduped; skip seq 5 → gap reported; unknown sandbox tolerated in v0.3 no-auth mode)
-
-### Wave 3: Sandbox telemetry wiring (depends on Wave 2)
-
-#### Task 2-3-01: Spawn sandboxes with the capture agent
-- **Persona:** sandbox-engineer — **REQ:** REQ-3-003
-- **Files:** `apps/ai-service/ai_service/sandbox/manager.py` (update), `apps/ai-service/ai_service/sandbox/unshare_backend.py` (update), `apps/ai-service/tests/sandbox/test_telemetry_wiring.py`
-- **Action:** `create()` gains optional `task_id`; when set the spawner copies `scripts/sandbox-agent.py` into the sandbox workdir, injects `NC_*` env, and launches the agent as a child of the namespace process (agent lifecycle tied to sandbox lifecycle; destroy kills the agent). No capture when task_id absent (pure shell sandbox).
-- **Verify:** `pnpm ai:test` — end-to-end on this box: create sandbox with task_id → run 2 commands via exec → events arrive at the ingest endpoint and land in SQLite in order
-
-### Wave 4: Reliability probe (depends on Wave 3)
-
-#### Task 2-4-01: Dropped-connection durability probe
-- **Persona:** backend-engineer — **REQ:** REQ-3-003
-- **Files:** `apps/ai-service/tests/telemetry/test_durability.py`
-- **Action:** Integration probe: run a capture agent against ingest, kill the WS connection mid-stream (simulate network failure), keep generating events, reconnect, assert the SQLite trace contains **every** event exactly once in order (spool + dedup). Document at-least-once semantics + replay path in README.
-- **Verify:** `pnpm ai:test` — test_durability green; README documents semantics
-
-### Must-Haves (Phase 2)
-- [ ] Real telemetry from a live sandbox arrives at ai-service: shell commands, file diffs, run/test results appear as ordered events in SQLite (end-to-end, no mocks)
-- [ ] Per-(learner,task) monotonic `seq`; gap detection reports missing seqs; replay yields the complete ordered trace
-- [ ] At-least-once proven: transient disconnect + reconnect loses no events; duplicates deduped server-side (durability probe green)
-- [ ] Capture agent is stdlib-only (AST-verified) and its lifecycle is tied to the sandbox (destroy kills it)
-- [ ] Trace retrievable by learner+task via `GET /v1/telemetry/traces/...`; unknown trace → 404
-- [ ] `TraceStore` protocol respected: no api/ code touches SQLite directly; `telemetry/` never imports `agents/` (D-027)
-- [ ] Flood control (G-3): burst past `AI_TELEMETRY_MAX_EVENTS_PER_TASK` → connection closed 1008 + trace marked `INCOMPLETE_FLOODED`; no silent event drop on overflow
-- [ ] `pnpm ai:test` green; `packages/types` telemetry TS types compile (`pnpm typecheck`)
-
----
-
-## Phase 3: Process-Trace Grading Engine
-
-**Requirements:** REQ-3-004
-**Goal:** Deterministic feature computation over traces (D-028) → compact digest → LLM rubric scoring via existing D-020 JSON defense → validated structured scores stored in `GradeStore`; LLM never sees the raw trace; engine calibrated against v0.2 mock corpora so process quality separates paste-and-run from iterative debugging
-
-### Wave 1: Features + grades store (parallel — no shared files)
-
-#### Task 3-1-01: Deterministic trace digest (features)
-- **Persona:** ai-engineer — **REQ:** REQ-3-004
-- **Files:** `apps/ai-service/ai_service/grading/__init__.py`, `apps/ai-service/ai_service/grading/features.py`, `apps/ai-service/tests/grading/__init__.py`, `apps/ai-service/tests/grading/test_features.py`
-- **Action:** Pure compute module (D-028): `compute_digest(trace: list[TelemetryEvent]) -> TraceDigest`. Deterministic features: test pass/fail counts + final status; edit count; error/fix cycle count + mean fix latency; idle gaps (>Ns, count + total); command category histogram (build/test/file/nav/debug/other); session duration; first-test-pass offset. `TraceDigest` pydantic model — compact (bounded size, no raw commands), LLM-safe.
-- **Verify:** `pnpm ai:test` — test_features green over synthetic traces: paste-and-run trace (0 error/fix cycles, single test pass at end) vs iterative trace (many cycles) produce observably different digests
-
-#### Task 3-1-02: GradeStore protocol + SQLite implementation
-- **Persona:** backend-engineer — **REQ:** REQ-3-004
-- **Files:** `apps/ai-service/ai_service/grading/store.py`, `apps/ai-service/tests/grading/test_store.py`
-- **Action:** `GradeStore` protocol (D-027): `save(grade)`, `get(learner_id, task_id)`, `list_for_learner(learner_id)`, `close()`. SQLModel `GradeRecord`: learner_id, task_id, variant_seed (null until P4), digest (JSON), scores (JSON), verdict, model, created_at. PK (learner_id, task_id). Postgres-migration-ready.
-- **Verify:** `pnpm ai:test` — test_store green (save/get/list roundtrip, overwrite-on-regrade documented)
-
-### Wave 2: Grading engine + calibration (depends on Wave 1)
-
-#### Task 3-2-01: Rubric scoring engine
-- **Persona:** ai-engineer — **REQ:** REQ-3-004
-- **Files:** `apps/ai-service/ai_service/grading/engine.py`, `apps/ai-service/ai_service/prompts/grading.py` (new), `apps/ai-service/tests/grading/test_engine.py`
-- **Action:** `GradingEngine.grade(learner_id, task_id) -> GradeRecord`: **trace-completeness gate (G-4)** — first call `TraceStore.gaps()` + check the trace `INCOMPLETE_FLOODED` flag; if gaps are non-empty OR the trace is flagged incomplete → return `verdict=UNGRADABLE_TRACE_INCOMPLETE` (a first-class verdict, not an exception) surfacing the gap list; a credential is NEVER issued from a gapped/incomplete trace. Otherwise: load trace via `TraceStore` → `compute_digest` → render rubric prompt (`prompts/grading.py`: criteria + level anchors for process quality, correctness, debugging discipline, test usage; a-4: treat high edit/command churn with no test-progress as a process-quality negative) → LLM structured output through the **existing D-020 4-layer defense** (`agents/structured.py` reused — engine composes it, never duplicates it) → validate `RubricScore` model (per-criterion 0-4 + strengths + gaps + verdict) → persist via `GradeStore`. Grading depends on `llm/` + `telemetry/` + `prompts/` only (boundary). Mock provider scripts deterministic rubric JSON for tests, including the INCOMPLETE path.
-- **Verify:** `pnpm ai:test` — test_engine green (mock provider: digest-only prompt asserted — **raw trace string absent from prompt**; validated scores returned; malformed JSON exercises D-020 retry; unknown trace → error)
-
-#### Task 3-2-02: Calibration against v0.2 mock corpora
-- **Persona:** ai-engineer — **REQ:** REQ-3-004
-- **Files:** `apps/ai-service/ai_service/corpus/trace_fixtures.py` (new), `apps/ai-service/tests/grading/test_calibration.py`
-- **Action:** Synthetic trace fixtures aligned with v0.2 `corpus/telemetry.py` + `corpus/artifacts.py` scenario IDs (strong/lazy/struggling builder archetypes). Assert grading separates them: strong archetype scores ≥ lazy archetype on process-quality criterion (mock provider maps digest shape → scripted scores; test asserts the ordering contract + that fixture IDs align with existing corpus IDs, D-021).
-- **Verify:** `pnpm ai:test` — test_calibration enforces the ordering contract
-
-### Wave 3: Grading endpoint (depends on Wave 2)
-
-#### Task 3-3-01: Assessment grade endpoint (real traces)
-- **Persona:** backend-engineer — **REQ:** REQ-3-004
-- **Files:** `apps/ai-service/ai_service/api/assessment.py` (update), `apps/ai-service/tests/api/test_grading.py`
-- **Action:** `POST /v1/assessment/grade {learner_id, task_id}` → `GradingEngine` → validated `RubricScore` JSON; `GET /v1/assessment/grade/{learner_id}/{task_id}` → stored grade; unknown trace → 404. Composed via DI (api/ owns wiring; engine knows nothing of FastAPI).
-- **Verify:** `pnpm ai:test` — test_grading green (grade roundtrip via TestClient with mock provider; 404 on unknown; GET after POST returns same scores)
-
-### Must-Haves (Phase 3)
-- [ ] Engine emits structured rubric-aligned scores from a **real process trace** (not pre-baked input) — TestClient roundtrip green
-- [ ] Deterministic features computed in code (test pass/fail, edit count, error/fix cycles, idle gaps, command categories); LLM receives the **digest only** — test asserts the raw trace never reaches the prompt (D-028)
-- [ ] Scores distinguish process quality: iterative-debugging archetype out-scores paste-and-run on the process criterion (calibration contract test)
-- [ ] Grades persisted + retrievable by learner+task via GradeStore (SQLite, protocol-wrapped, D-027)
-- [ ] Boundary rules hold: `grading/` imports no `api/`/`agents/` internals except the shared D-020 structured defense; `pnpm ai:test` + `pnpm ai:lint` green
-- [ ] Incomplete-trace gate (G-4): gapped or `INCOMPLETE_FLOODED` trace → `verdict=UNGRADABLE_TRACE_INCOMPLETE` with the gap list; no credential issued from an incomplete trace (test green)
-
----
-
-## Phase 4: Variant Task Generation
-
-**Requirements:** REQ-3-005
-**Goal:** Template library with typed parameter slots (D-029) + seeded LLM instantiation + per-learner variant registry (SQLite) with difficulty-normalization anchors; two learners on the same competency get provably distinct, reproducible, auditable tasks
-
-### Wave 1: Templates + store (parallel — no shared files)
-
-#### Task 4-1-01: Task template library
-- **Persona:** ai-engineer — **REQ:** REQ-3-005
-- **Files:** `apps/ai-service/ai_service/variants/__init__.py`, `apps/ai-service/ai_service/variants/templates.py`, `apps/ai-service/tests/variants/__init__.py`, `apps/ai-service/tests/variants/test_templates.py`
-- **Action:** ≥3 initial task templates bound to existing competency IDs (D-021 alignment). `TaskTemplate`: id, competency_id, statement skeleton with `{slot}` placeholders, `ParameterSlot[]` (name, type: enum/int-range/string-set, allowed values), `rubric anchors` (difficulty normalization: expected feature envelope — e.g. expected edit-count band — used by grading context), starter-file scaffolds served to the sandbox. Seeded slot sampler is pure code (`random.Random(seed)`), fully reproducible.
-- **Verify:** `pnpm ai:test` — test_templates green (slot validation: bad value rejected; seeded sampling reproducible across runs; all templates bind to real competency IDs)
-
-#### Task 4-1-02: VariantStore protocol + SQLite implementation
-- **Persona:** backend-engineer — **REQ:** REQ-3-005
-- **Files:** `apps/ai-service/ai_service/variants/store.py`, `apps/ai-service/tests/variants/test_store.py`
-- **Action:** `VariantStore` protocol (D-027): `save(variant)`, `get(learner_id, template_id_or_task_id)`, `list_for_learner(learner_id)`, `list_by_template(template_id)`, `close()`. SQLModel `VariantRecord`: learner_id, task_id (the grading/telemetry task key), template_id, seed, params (JSON), statement (rendered), created_at. Unique (learner_id, template_id).
-- **Verify:** `pnpm ai:test` — test_store green (roundtrip, unique constraint, audit listing)
-
-### Wave 2: Generator (depends on Wave 1)
-
-#### Task 4-2-01: Seeded LLM variant generator
-- **Persona:** ai-engineer — **REQ:** REQ-3-005
-- **Files:** `apps/ai-service/ai_service/variants/generator.py`, `apps/ai-service/ai_service/prompts/variant.py` (new), `apps/ai-service/tests/variants/test_generator.py`
-- **Action:** `VariantGenerator.generate(learner_id, template_id) -> VariantRecord`: derive seed (`sha256(template_id|learner_id|milestone)` — reproducible, D-029); sample typed slots in code; render a fill prompt (statement skeleton + concrete slot values) → LLM via D-020 structured defense → unique task statement + starter files → validate → persist (seed + params + statement) via `VariantStore`. Cache: existing (learner,template) returns the stored variant (no duplicate work). Mock provider scripts deterministic statements per seed for tests.
-- **Verify:** `pnpm ai:test` — test_generator green: two different learner_ids → distinct statements for the same template; same learner twice → identical stored variant (reproducible); params JSON contains only schema-valid slot values; **fairness envelope (a-5):** two variants of one template compute digests within the template's expected feature envelope (comparable slot complexity/difficulty features) — "same bar" is testable, not asserted
-
-### Wave 3: Variant endpoint + TS types (depends on Wave 2)
-
-#### Task 4-3-01: Variant task endpoint
-- **Persona:** backend-engineer — **REQ:** REQ-3-005
-- **Files:** `apps/ai-service/ai_service/api/variants.py` (new), `apps/ai-service/ai_service/main.py` (update), `apps/ai-service/tests/api/test_variants.py`
-- **Action:** `POST /v1/variants {learner_id, template_id or competency_id}` → generated (or cached) variant: statement, starter files, task_id, seed; `GET /v1/variants/{task_id}` → stored variant; `GET /v1/variants?learner_id=` → learner's variants. DI wiring in api/ only.
-- **Verify:** `pnpm ai:test` — test_variants green (generate→get roundtrip; cache hit on regenerate; distinct learners → distinct statements asserted at the API layer)
-
-#### Task 4-3-02: TS types for variants (+ grades)
-- **Persona:** data-engineer — **REQ:** REQ-3-005
-- **Files:** `packages/types/variants.ts` (new), `packages/types/grading.ts` (new), `packages/types/index.ts` (update)
-- **Action:** TS `TaskVariant`, `VariantParams`, `RubricScore`, `GradeRecord` matching Python models (cross-referencing headers; same field names). Consumed by Phase 6 surfaces.
-- **Verify:** `pnpm typecheck` passes
-
-### Must-Haves (Phase 4)
-- [ ] Two learners requesting the same competency receive **provably distinct** task variants (API-level test)
-- [ ] Seed derivation reproducible: same (template, learner) → same variant, served from cache without a second LLM call (D-029)
-- [ ] Variant seed + typed params persisted + auditable (VariantStore listing; proctoring cross-check path exists)
-- [ ] Difficulty normalization anchors present per template and shipped to the grader prompt context
-- [ ] Starter-file scaffolds defined per template (P6 wires them into the sandbox workdir)
-- [ ] `pnpm ai:test` + `pnpm typecheck` green; `variants/` imports no api/ (boundary)
-
----
-
-## Phase 5: Oral / Voice Defense
-
-**Requirements:** REQ-3-006
-**Goal:** `VoiceProvider` protocol with mock + browser fallback (D-030) + seventh `Examiner` agent streaming over existing SSE + `DefenseStore` persisting transcript + integrity signals. **Real server STT/TTS (`OpenAIAudioProvider`) is DEFERRED to v0.4 (with KYC, when there's a real key + real users)** — voice is mock-first (D-030) and the `/audio/*` real path could never be exercised in CI, so v0.3 proves the full defense *dialogue* + integrity-signal pipeline over mock + browser-native fallback only; the protocol seam keeps the real provider a drop-in later.
-
-### Wave 1: Voice provider layer (parallel — no shared files)
-
-#### Task 5-1-01: VoiceProvider protocol + mock provider + browser fallback
-- **Persona:** voice-engineer — **REQ:** REQ-3-006
-- **Files:** `apps/ai-service/ai_service/voice/__init__.py`, `apps/ai-service/ai_service/voice/base.py`, `apps/ai-service/ai_service/voice/mock.py`, `apps/ai-service/ai_service/voice/browser.py`, `apps/ai-service/ai_service/voice/factory.py`, `apps/ai-service/tests/voice/__init__.py`, `apps/ai-service/tests/voice/test_mock.py`, `apps/ai-service/tests/voice/test_factory.py`
-- **Action:** `VoiceProvider` protocol mirroring `LLMProvider` (D-030): `transcribe(audio: bytes, fmt) -> TranscriptSegment` + `synthesize(text, voice) -> AsyncIterator[bytes]`. `MockVoiceProvider`: deterministic canned transcript (scripted per test), canned 1kHz-tone WAV bytes, scripted failure modes. `browser.py`: fallback **descriptor** (`sr_available: true`, endpoint hints) the web client uses to select browser-native `SpeechRecognition`/`speechSynthesis` when no server provider. `factory.py`: `AI_VOICE_PROVIDER=browser | mock` (default mock when no key). **`OpenAIAudioProvider` (real server STT/TTS) intentionally NOT built in v0.3 — deferred to v0.4**; the protocol is its future seam. `voice/` never imports `agents/` or `api/`.
-- **Verify:** `pnpm ai:test` — test_mock + test_factory green (deterministic transcribe/synthesize; failure modes; factory selects mock with empty key, browser when provider=browser; zero network calls)
-
-#### Task 5-1-03: DefenseStore protocol + SQLite implementation
-- **Persona:** backend-engineer — **REQ:** REQ-3-006
-- **Files:** `apps/ai-service/ai_service/voice/defense_store.py`, `apps/ai-service/tests/voice/test_defense_store.py`
-- **Action:** `DefenseStore` protocol (D-027): `start(defense)`, `append_turn(defense_id, turn)`, `finalize(defense_id, integrity_signals)`, `get(defense_id)`, `list_for_learner(learner_id)`, `close()`. SQLModel `DefenseRecord` (id, learner_id, task_id, status, created/finished_at) + `DefenseTurn` (defense_id FK, turn seq, role examiner|learner, text, ts, latency_ms) + integrity signals JSON on the record (long pauses, off-scope cadence markers — A-109).
-- **Verify:** `pnpm ai:test` — test_defense_store green (start→append turns→finalize→get roundtrip; ordered turns by seq)
-
-### Wave 2: Examiner agent (depends on Wave 1)
-
-#### Task 5-2-01: Examiner agent (seventh agent)
-- **Persona:** ai-engineer — **REQ:** REQ-3-006
-- **Files:** `apps/ai-service/ai_service/agents/examiner.py`, `apps/ai-service/ai_service/prompts/examiner.py` (new), `apps/ai-service/ai_service/agents/registry.py` (update: central registration, G-4 pattern), `apps/ai-service/tests/agents/test_examiner.py`
-- **Action:** `ExaminerAgent(BaseAgent)` (D-030/A-109): builds questions from learner transcript + trace digest + (P4) variant statement; probes understanding + challenges process choices ("why did you choose X at step N?"); streams questions over the existing SSE pipeline; `structured` verdict mode returns verdict + per-answer integrity signal list (long pause flags, off-scope answers) computed from turn metadata; calls voice **only through the `VoiceProvider` protocol** (PERSONAS conflict rule — never concrete providers). Session-scoped history reused from v0.2.
-- **Verify:** `pnpm ai:test` — test_examiner green (question stream references trace-digest facts; verdict structured output validates via D-020 defense; registry resolves all seven agents; mock-VoiceProvider wiring through protocol only — asserted by import scan in test)
-
-### Wave 3: Defense endpoints (depends on Wave 2)
-
-#### Task 5-3-01: Defense session + audio endpoints
-- **Persona:** backend-engineer — **REQ:** REQ-3-006
-- **Files:** `apps/ai-service/ai_service/api/defense.py` (new), `apps/ai-service/ai_service/main.py` (update), `apps/ai-service/tests/api/test_defense.py`
-- **Action:** `POST /v1/defense/start {learner_id, task_id}` → creates DefenseRecord + streams the first examiner question (SSE, agent=examiner); `POST /v1/defense/{id}/answer` (multipart audio from browser MediaRecorder, or `{text}` for typed fallback) → STT via VoiceProvider → append learner turn → stream examiner follow-up (SSE) → TTS audio chunks over `GET /v1/defense/{id}/audio/{turn_id}`; `POST /v1/defense/{id}/finish` → verdict + integrity signals persisted; `GET /v1/defense/{id}` → full transcript + signals. Browser-fallback mode: when provider=browser, start returns the fallback descriptor instead of server audio.
-- **Verify:** `pnpm ai:test` — test_defense green (full loop with mock voice + mock LLM: start → answer(text) → answer(audio bytes) → finish → transcript retrievable with per-turn latency; unknown id → 404; `GET` signals present after finish)
-
-### Wave 4: Examiner latency instrumentation (depends on Wave 3)
-
-#### Task 5-4-01: Per-turn latency instrumentation (mock-based)
-- **Persona:** voice-engineer — **REQ:** REQ-3-006
-- **Files:** `apps/ai-service/tests/voice/test_latency.py`, `apps/ai-service/README.md` (update: conversational-budget doc + v0.4 voice note)
-- **Action:** Instrument per-turn latency (STT ms + LLM TTFT ms + TTS ms) recorded on each DefenseTurn; deterministic test over mock providers asserts instrumentation presence + `latency_ms` populated + budget constant defined (mock runs are near-instant — wall-clock asserted in v0.4 against a real endpoint). README documents the conversational-latency target as a v0.4 acceptance criterion (real STT/TTS deferred per CUT-1/G-7).
-- **Verify:** `pnpm ai:test` — test_latency green (latency_ms fields populated on every turn; budget constant defined); README documents the deferred real-voice acceptance probe
-
-### Must-Haves (Phase 5)
-- [ ] Spoken defense runs end-to-end over HTTP with mock providers: start → answer (audio + typed fallback) → examiner follow-up streams → verdict + transcript persisted (automated)
-- [ ] Examiner is the seventh registered agent; streams over the existing SSE envelope (meta agent=examiner)
-- [ ] Instrumented per-turn latency fields populated on every DefenseTurn (STT ms + LLM TTFT ms + TTS ms); conversational budget named (A-109)
-- [ ] `VoiceProvider` protocol respected: examiner + api touch voice only via the protocol; mock-first — **no task requires a real voice key to pass**
-- [ ] Browser-native SR/TTS fallback descriptor returned when no server voice provider configured (mock/browser are first-class, D-030)
-- [ ] Real server STT/TTS (`OpenAIAudioProvider`) explicitly deferred to v0.4 (with real keys/users); the defense pipeline is fully proven over mock+browser — documented in README + release note
-- [ ] Optional future voice config noted for v0.4 (`AI_VOICE_BASE_URL` / `AI_VOICE_API_KEY`) in `.env.example` + README; keys only in gitignored `.ciagent/.env.secrets`; tests never call a voice API
-- [ ] Boundary rules hold: `voice/` imports no `agents/`/`api/`; `pnpm ai:test` + `pnpm ai:lint` green
-
----
-
-## Phase 6: Agent Re-grounding + Learner Surface Integration
-
-**Requirements:** REQ-3-007, REQ-3-008
-**Goal:** Lab/Assessor/Proctor consume real engine inputs (live telemetry, grading output, defense signals) with **no mock fallback in the learner path**; the v0.1 sandbox + assessment mockups become real — in-browser build/run (Run/Test buttons + read-only exec output, CUT-2 — no interactive shell), live telemetry panel, browser/typed voice defense, live grading; `pnpm build` + `pnpm typecheck` + full `pnpm ai:test` green
-**Note:** E2E verification runs against the real engines over HTTP with `AI_PROVIDER=mock` + `AI_VOICE_PROVIDER=mock` permitted (G-2 precedent) — the requirement is real engine plumbing (sandbox/telemetry/grading/defense over real endpoints, no corpus mocks in the learner path); a cloud outage must not block P6.
-
-### Wave 1: Agent re-grounding (parallel — no shared files)
-
-#### Task 6-1-01: Lab agent on live telemetry
-- **Persona:** ai-engineer — **REQ:** REQ-3-007
-- **Files:** `apps/ai-service/ai_service/agents/lab.py` (update), `apps/ai-service/ai_service/prompts/lab.py` (update: render real trace digest), `apps/ai-service/tests/agents/test_lab_live.py` (new)
-- **Action:** Lab consumes a **live trace digest** (grading/features `compute_digest` over `TraceStore` events) instead of `corpus/telemetry.py`. build_messages renders digest facts (recent commands, failing tests, idle). Mock-provider scripts assert digest-derived content. v0.2 corpus path removed from the agent (dormant corpus retained until Task 6-1-04 check).
-- **Verify:** `pnpm ai:test` — test_lab_live green: feedback references events actually present in a seeded SQLite trace (not corpus fixtures); no `corpus.telemetry` import in `agents/lab.py` (AST-asserted)
-
-#### Task 6-1-02: Assessor agent on grading output
-- **Persona:** ai-engineer — **REQ:** REQ-3-007
-- **Files:** `apps/ai-service/ai_service/agents/assessor.py` (update), `apps/ai-service/ai_service/prompts/assessor.py` (update), `apps/ai-service/tests/agents/test_assessor_live.py` (new)
-- **Action:** Assessor consumes `GradeStore` output (validated `RubricScore` + digest) for learner+task instead of pre-baked artifacts/transcripts; renders strengths/gaps/verdict with rubric-anchored coaching framing. Structured output unchanged (D-020).
-- **Verify:** `pnpm ai:test` — test_assessor_live green: given a real stored grade, Assessor output reflects its scores; corpus artifact path gone from the agent (AST-asserted)
-
-#### Task 6-1-03: Proctor agent on telemetry + defense signals
-- **Persona:** ai-engineer — **REQ:** REQ-3-007
-- **Files:** `apps/ai-service/ai_service/agents/proctor.py` (update), `apps/ai-service/ai_service/prompts/proctor.py` (update), `apps/ai-service/tests/agents/test_proctor_live.py` (new)
-- **Action:** Proctor consumes real integrity inputs: idle gaps + command cadence from the trace digest + defense integrity signals from `DefenseStore` → classified signals + coaching interventions (supportive tone retained). Cross-checks variant seed params (P4) for off-template work.
-- **Verify:** `pnpm ai:test` — test_proctor_live green: signals derived from seeded real trace + defense records; corpus proctor scenarios no longer imported (AST-asserted)
-
-#### Task 6-1-04: Corpus dormancy + mockup removal verification
-- **Persona:** lead-developer — **REQ:** REQ-3-007
-- **Files:** `apps/ai-service/ai_service/corpus/telemetry.py`, `apps/ai-service/ai_service/corpus/artifacts.py`, `apps/ai-service/ai_service/corpus/` (README note), `apps/ai-service/tests/test_corpus_dormancy.py` (new)
-- **Action:** Verify no production code path imports `corpus/telemetry.py` or `corpus/artifacts.py` anymore (test scans imports across `agents/`, `api/`, engines). Retain files as Phase-3 calibration history with a header note marking them **dormant — v0.2 mocks, not used at runtime**; learner-context corpus stays (agents still need learner context). Disposes v0.2's G-5-class dead-code risk deliberately.
-- **Verify:** `pnpm ai:test` — test_corpus_dormancy green (zero runtime importers); suite otherwise unchanged
-
-### Wave 2: Client plumbing + design primitives (parallel — no shared files)
-
-#### Task 6-2-01: Sandbox build-panel engine client (run/test-only — no raw shell relay)
-- **Persona:** frontend-engineer — **REQ:** REQ-3-008
-- **Files:** `apps/web/hooks/use-sandbox-session.ts` (new), `apps/web/lib/engine-client.ts` (new), `apps/web/.env.example` (update)
-- **Action:** `engine-client.ts`: typed fetch client for `/v1/sandboxes` (create/destroy — learner_id from the v0.3 mock session constant, allowlisted server-side per G-5), `/v1/sandboxes/{id}/files` (workspace CRUD), `/v1/sandboxes/{id}/exec` (run/test), `/v1/variants`, `/v1/assessment/grade`, `/v1/telemetry/traces`, `/v1/defense/*`; base `NEXT_PUBLIC_AI_SERVICE_URL`. `use-sandbox-session.ts`: create sandbox+variant on task open → destroy on unmount (idempotent cleanup, AbortController pattern); 503 pool-full → user-facing "environment busy, retry" (D-032); 403/429 abuse-control surfaced honestly (G-5). **CUT-2 (G-8): NO raw interactive WS terminal relay (keystroke-level stdin/stdout) in v0.3** — the credential pipeline needs *process events* (from Run/Test + file edits), not a live shell; the interactive xterm relay is the most fragile real-time piece and is deferred to v0.4. The build panel is a **Run/Test output viewer** (exec results + telemetry pulse render), not an interactive shell. `@xterm/xterm` is therefore NOT a dependency in v0.3.
-- **Verify:** `pnpm install && pnpm typecheck` pass; hook unmount destroys the sandbox (manual probe: `curl localhost:8420/v1/sandboxes` shows count drop after navigation); RUN/TEST buttons produce streamed output + telemetry events in the trace
-
-#### Task 6-2-02: New design primitives
-- **Persona:** design-system-engineer — **REQ:** REQ-3-008
-- **Files:** `packages/ui/src/primitives/terminal-frame.tsx` (new), `packages/ui/src/primitives/mic-control.tsx` (new), `packages/ui/src/primitives/grade-badge.tsx` (new), `packages/ui/src/primitives/telemetry-status.tsx` (new), `packages/ui/src/primitives/transcript-viewer.tsx` (new), `packages/ui/src/primitives/index.ts` (update), `packages/ui/src/index.ts` (update)
-- **Action:** Token-driven primitives: TerminalFrame (CUT-2: a read-only exec-output viewer chrome — streams Run/Test results, NOT an interactive shell), MicControl (record/stop with consent state + no-mic fallback state, MediaRecorder permission UX), GradeBadge (verdict rendering), TelemetryStatus (live event pulse / disconnected indicator), TranscriptViewer (examiner/learner turn list). Dark mode + WCAG AA; stories for each.
-- **Verify:** primitives import from `@nextcraft/ui`; Storybook stories render dark + light; `pnpm build` (ui package) passes
-
-#### Task 6-2-03: Defense TS types
-- **Persona:** data-engineer — **REQ:** REQ-3-008
-- **Files:** `packages/types/defense.ts` (new), `packages/types/index.ts` (update)
-- **Action:** TS `DefenseSession`, `DefenseTurn`, `IntegritySignal`, `Verdict` mirroring P5 Python models (cross-referencing header).
-- **Verify:** `pnpm typecheck` passes
-
-### Wave 3: Real build surface (depends on Wave 2)
-
-#### Task 6-3-01: Sandbox mockup → real in-browser IDE
-- **Persona:** frontend-engineer — **REQ:** REQ-3-008
-- **Files:** `apps/web/app/(learner)/build/[competencyId]/page.tsx` (rewrite), `apps/web/components/learner/sandbox-terminal.tsx` (new), `apps/web/components/learner/file-tree.tsx` (new), `apps/web/components/learner/run-controls.tsx` (new), `apps/web/components/learner/lab-feedback-panel.tsx` (update: live trace)
-- **Action:** Replace the mockup with the real build environment (A-103, CUT-2): file tree (HTTP CRUD into the sandbox workdir via `/v1/sandboxes/{id}/files` routes added to api/sandboxes — read/write/list workspace files), syntax-highlight editor (existing), **Run**/**Test** buttons (exec in sandbox; results stream to a read-only TerminalFrame output panel — no interactive shell), starter files from the P4 variant scaffold. Lab panel posts `learner_id+task_id` → streams Lab feedback over the **live** trace (no scenario IDs). Telemetry sidebar shows live TelemetryStatus. Pool-full 503 → busy state with retry; 403/429 surfaced.
-- **Verify:** with ai-service up: open `/build/comp-01` → variant statement + starter files load → edit a file → **Run** executes the command in-sandbox and output renders in the panel → **Test** runs the test suite in-sandbox → Lab panel streams digest-derived feedback → telemetry status shows live events. Manual probe documented; `pnpm typecheck` green
-
-### Wave 4: Live defense + grading surfaces (depends on Waves 2-3)
-
-#### Task 6-4-01: Assessment mockup → live defense + live grading
-- **Persona:** frontend-engineer — **REQ:** REQ-3-008
-- **Files:** `apps/web/app/(learner)/defend/[competencyId]/page.tsx` (rewrite), `apps/web/components/learner/defense-session.tsx` (new), `apps/web/components/learner/assessor-results-panel.tsx` (update), `apps/web/components/learner/proctor-banner.tsx` (update), `apps/web/components/learner/oral-defense-interface.tsx` (rewrite or remove)
-- **Action:** Real assessment flow: **Start Defense** → POST `/v1/defense/start` → examiner question streams → learner answers via MicControl (MediaRecorder webm/opus → multipart POST) with typed fallback when mic denied or `provider=browser` (native `SpeechRecognition`/`speechSynthesis` path per fallback descriptor) → follow-ups stream → **Finish** → verdict + integrity signals panel (TranscriptViewer, GradeBadge) + **Grade My Work** → POST `/v1/assessment/grade` → structured rubric bars render. Proctor banner reads signals for task_id. Loading + error states throughout; no mock defense data remains in the learner path.
-- **Verify:** with ai-service up: full defense loop runs in-browser (typed fallback acceptable in CI-less manual probe; mic path exercised manually with permission granted); grade panel renders real rubric scores; `pnpm typecheck` green
-
-### Wave 5: End-to-end verification (depends on Waves 3-4)
-
-#### Task 6-5-01: Full learner-path E2E probe + green builds
-- **Persona:** backend-engineer — **REQ:** REQ-3-007, REQ-3-008
-- **Files:** `apps/ai-service/tests/api/test_e2e_credential_flow.py` (new), `apps/ai-service/README.md` (update: E2E probe doc)
-- **Action:** Endpoint-level E2E (mock LLM/voice providers, real engines): create variant → create sandbox with task_id → ingest trace events via real sandbox exec → POST grade → start defense (typed answers) → finish → assert: trace persisted, grade stored + digest-linked, defense transcript + signals stored, Proctor/Assessor endpoints serve them. No corpus fixture anywhere in the flow (AST-asserted). Run `pnpm build` + `pnpm typecheck` + full `pnpm ai:test` at repo root; fix all failures before phase ship.
-- **Verify:** `pnpm ai:test` green incl. test_e2e_credential_flow; `pnpm build` + `pnpm typecheck` green; README E2E probe section documents the manual browser pass
-
-### Must-Haves (Phase 6)
-- [ ] Lab/Assessor/Proctor operate on real inputs with **no mock fallback in the learner path** (AST-verified: no corpus telemetry/artifact/proctor imports in production paths)
-- [ ] Learner builds in-browser for real (CUT-2): Run/Test buttons execute in a namespace sandbox and stream output to a read-only panel; file tree CRUD works; starter files come from the learner's variant scaffold
-- [ ] Live telemetry: build activity streams to ai-service and the sidebar shows live status (TelemetryStatus); trace persisted in SQLite
-- [ ] Live defense: start → answer (mic or typed fallback) → examiner follow-ups → finish → transcript + integrity signals + verdict rendered; browser-native path works with no server voice key
-- [ ] Live grading: grade request returns structured rubric scores computed from the real trace digest; grade panel renders them
-- [ ] 503 pool-full surfaced honestly in UI; navigating away destroys the sandbox (no leaked sandboxes — `GET /v1/sandboxes` manual probe)
-- [ ] Examiner remains protocol-clean (voice only via `VoiceProvider`); module boundary rules hold across all new code
-- [ ] `pnpm build` and `pnpm typecheck` pass; full `pnpm ai:test` green; no cloud/voice calls in any automated test
-- [ ] Release-note input (for P7): v0.3 ships real engines; **identity/age-gating (KYC) remains deferred — age-gating is still a visual mockup** (A-110); sandbox scope is coding-IDE only (design tool/simulation deferred to v0.4, D-025)
-
----
-
-## Phase 7: Final Review + Ship (no planned tasks)
-
-Orchestrated by the SHIP stage, not this plan: multi-persona code review (correctness, testing, module boundaries, secrets hygiene — keys absent from code/logs/commits/errors, localhost-only CORS, no PII in prompts), project health audit (reconstruction, .ciagent/ discipline, branch/commit hygiene), then merge milestone → main, tag the final v0.2.x patch, create the Gitea release, mark all 8 v0.3 requirements complete.
-
-**Release-note honesty:** the release note must state (a) Lab/Assessor/Proctor now run on real engine inputs (v0.2 mock-input caveat retired), (b) sandbox scope = coding IDE only — design tool + simulation environments deferred to v0.4 (D-025), (c) identity/age-gating (KYC) deferred per founder directive — age-gating remains the v0.1 visual flow mockup; abuse control (per-learner sandbox caps + server-side learner allowlist, G-5) ships in place of auth (A-110), (d) voice runs mock-first with browser-native fallback — real server STT/TTS deferred to v0.4 (CUT-1), (e) sandbox resource limits are partially enforced (memory/CPU/wall-clock kernel-enforced via rlimits; per-sandbox pids + hard disk quota are NOT — mitigated by a workdir-size sweep and per-learner caps; full enforcement requires cgroup delegation, deferred to the post-MVP containerd backend, D-024/G-1/G-2).
-
-**Secrets-hygiene checklist (P7):** `.ciagent/.env.secrets` gitignored and never committed; `AI_VOICE_API_KEY` / `AI_TUTOR_API_KEY` referenced only via env; SQLite DBs + sandbox dirs gitignored; no keys in logs, error messages, or test fixtures.
-
-**Disposal checks (G-5 class):** v0.2 dormant corpus files carry the dormant-header note (Task 6-1-04); any now-unused mock-data exports for the old sandbox/defense mockups (e.g. `aiTutorResponses`-class leftovers) must be removed or deprecated by review.
-
----
+| 1 | Bootstrap CLI core | REQ-4-001, REQ-4-002 | 3 | cli-engineer, backend-engineer, security-auditor (W3 review) |
+| 2 | Binary build + release pipeline | REQ-4-003, REQ-4-004 | 3 | cli-engineer, backend-engineer, security-auditor |
+| 3 | Install docs + fresh-clone E2E | REQ-4-005 | 2 | cli-engineer, backend-engineer |
+| 4 | Final review + ship | — | 1 | all reviewers |
 
 ## User-Facing Surface
 
-The primary user-facing surface is the **learner build + defend flow** at `http://localhost:3000`, backed by the real engines in ai-service at `http://localhost:8420`:
-
-- `/dashboard` — AI tutor chat (Coach/Tutor, streaming) + Mentor panel (unchanged from v0.2)
-- `/learn/[competencyId]` — byte viewer with streaming Tutor explanations (unchanged)
-- `/build/[competencyId]` — **real in-browser build environment**: file tree, syntax editor, **Run/Test buttons that execute in a namespace sandbox and stream output to a read-only panel** (CUT-2 — no interactive shell), live telemetry status, live Lab feedback, per-learner variant task statement
-- `/defend/[competencyId]` — **live oral defense + live grading**: Examiner voice/typed dialogue, transcript + integrity signals, real rubric scores from the process trace
-
-The marketplace, employer, and admin surfaces are unchanged from v0.1/v0.2.
+1. **One-liner install (README quickstart):** `curl -fsSL https://git.coreci.dev/coreci/nextcraft/raw/main/scripts/install.sh | bash` — downloads the latest release's `nextcraft-linux-x64` binary, verifies its sha256, installs to `~/.local/bin`, prints a PATH hint if needed.
+2. **CLI commands:** `nextcraft doctor` (prereq checks), `nextcraft bootstrap` (fresh clone → runnable stack), `nextcraft verify` (health check), `nextcraft dev` (dev server passthrough), plus `--help`/`--version`.
+3. **Release surface:** every Gitea release from v0.3.2 onward carries `nextcraft-linux-x64` + `nextcraft-linux-x64.sha256` assets.
 
 ## Happy Path
 
-1. Learner opens `/build/comp-01` → a per-learner **variant statement** and starter files load; a namespace sandbox is created for the session
-2. Learner edits files in the tree and clicks **Run**/**Test** → commands execute in the sandbox and real output renders in the build panel; the telemetry sidebar pulses as events stream to ai-service and persist in SQLite
-3. The Lab panel streams feedback derived from the **live trace digest** (real commands, real failures)
-4. Learner opens `/defend/comp-01` → **Start Defense**: the Examiner streams an opening question ("Walk me through your build — why did you structure it this way?")
-5. Learner answers by voice (mic consent → MediaRecorder → STT) or typed fallback → examiner follow-ups probe the trace ("You hit three test failures before passing — what changed?"); TTS plays examiner audio (or browser speech in fallback)
-6. Learner finishes the defense → transcript + integrity signals appear; verdict renders in a GradeBadge
-7. Learner clicks **Grade My Work** → the grading engine computes the digest from the real trace, rubric-scores it, and the panel renders per-criterion bars + strengths/gaps/verdict
-8. Proctor banner shows integrity signals from the live trace + defense in coaching tone; Mentor panel on `/dashboard` can narrate the real outcome
-9. Navigating away destroys the sandbox (pool slot freed); killing ai-service shows inline error + retry states on every panel, with no crashes
+Before execution, the end-to-end scenario this milestone must make true:
+
+1. A consumer on a linux x64 box runs the one-liner; `nextcraft` lands in `~/.local/bin`.
+2. They clone the repo (or the CLI detects the repo root), run `nextcraft doctor` — all prerequisites report ✓ with actionable messages for any gap.
+3. `nextcraft bootstrap` — pnpm install, ai-service venv via the existing bootstrap.sh, `.env` created from `.env.example`, optional-key warnings (not blockers), mock providers keep the stack runnable keyless.
+4. `nextcraft verify` — venv imports, ports, env presence, build readiness all ✓.
+5. `nextcraft dev` — the dev stack runs; Ctrl+C stops it (passthrough semantics).
+6. On every ship, the Gitea release shows the binary + checksum assets; re-running the one-liner upgrades to the latest binary.
 
 ## UX Acceptance Criteria
 
-1. Run/Test output visibly reflects the real sandbox execution (command round-trip to the sandbox, real stdout/stderr), not a replay animation
-2. The learner path contains **no mock engine data** — scenarios, canned artifacts, and scripted defense transcripts from v0.2 are gone from runtime
-3. Variant statements visibly differ between two learner sessions on the same competency
-4. Mic permission flow is graceful: consent prompt, recording indicator, no-mic/typed fallback, and browser-native speech path when no server voice key is configured
-5. Defense transcript renders turn-by-turn with latency shown; integrity signals render in coaching (supportive) tone
-6. Grade results render as structured per-criterion bars with verdict, from the real trace — not from final-output-only heuristics
-7. Pool-full (503) shows an honest "environment busy — retry" state; navigation/unmount destroys sandboxes with no leaks
-8. When ai-service is unreachable: inline error + retry on every panel — no crashes, no console errors, no blank UI
-9. All new UI uses design tokens, supports dark mode, meets WCAG AA contrast, responsive at 375px / 768px / 1280px
-10. `pnpm build` and `pnpm typecheck` pass with zero errors; `pnpm ai:test` green cloud-free and voice-key-free
+- `doctor` output lists every prerequisite with ✓/✗ and a **fix hint** on every ✗; exit code 1 if any ✗, 0 otherwise.
+- `bootstrap` is **idempotent** — running twice produces the same end state, second run fast (no reinstalls where avoidable).
+- `bootstrap` never writes secrets, never blocks on missing optional keys — warns with the exact key names and where to set them.
+- `verify` gives a single-glance green/red summary; every red item names the failing command it ran.
+- Every command supports `--help`; unknown command/flag exits 2 with usage.
+- The one-liner **never hard-fails silently**: any error path (no release, no binary asset, checksum mismatch, platform mismatch) prints a specific message + the source-bootstrap alternative.
+- Checksum mismatch = hard stop + explicit "do not run this binary" message.
+- PATH hint: if `~/.local/bin` is not on PATH, the installer prints the exact export line to add.
+- Binary runs standalone on a box with node NOT installed (SEA self-containment) — `./nextcraft-linux-x64 --version` works.
+
+---
+
+## Phase 1: Bootstrap CLI Core
+
+**Requirements:** REQ-4-001, REQ-4-002
+**Goal:** `apps/cli` package with doctor/bootstrap/verify/dev fully working from source (`node dist` + pnpm bin), unit-tested, wired into the monorepo (turbo + root scripts), composing — not duplicating — the existing scripts.
+
+### Wave 1: Package foundation (parallel)
+
+#### Task 1-1-01: CLI package scaffold + entry + dispatch
+- **Persona:** cli-engineer — **REQ:** REQ-4-001
+- **Files:** `apps/cli/package.json`, `apps/cli/tsconfig.json`, `apps/cli/src/index.ts`, `apps/cli/src/commands/help.ts` (usage text), `apps/cli/tests/dispatch.test.ts`
+- **Action:** pnpm workspace package `@nextcraft/cli` (private, `"bin": {"nextcraft": "dist/index.js"}`). Entry: parse argv (hand-rolled, no runtime deps), dispatch to commands, `--help`/`-h`, `--version` (from package.json version), unknown → exit 2 with usage. Exit-code contract: 0 ok / 1 failure / 2 usage. shebang `#!/usr/bin/env node` on the built entry (esbuild banner in P2; for P1 `tsx` runs in dev via package script `"dev": "tsx src/index.ts"`).
+- **Verify:** `pnpm --filter @nextcraft/cli test` green (dispatch: routes doctor/bootstrap/verify/dev; unknown exits 2; --help exits 0; --version prints package version); `pnpm typecheck` green.
+
+#### Task 1-1-02: Checks library (pure logic)
+- **Persona:** cli-engineer — **REQ:** REQ-4-001, REQ-4-002
+- **Files:** `apps/cli/src/checks/check-command.ts`, `apps/cli/src/checks/check-env.ts`, `apps/cli/src/lib/log.ts`, `apps/cli/tests/checks.test.ts`
+- **Action:** `check-command`: given a name + optional `--version` probe + a min-version parser, resolve binary on PATH (`which`), semver-ish compare (major.minor tolerant), return `CheckResult {name, ok, found, version, hint}`. `check-env`: diff `.env.example` template keys vs an existing `.env` (missing keys → warn-classified; required-vs-optional classification table from the template's own comments + a static required list of zero keys — all optional per A-210), return per-key results. `log.ts`: `ok(msg)`, `fail(msg, hint)`, `warn(msg)`, `info(msg)` formatters with symbols and consistent alignment. Pure functions — no side effects at import; fs access injected as parameters for testability.
+- **Verify:** unit tests green: version compare (>= boundaries), missing binary → ok:false + hint, env diff missing/new/extra keys, required-optional classification.
+
+#### Task 1-1-03: Root + turbo wiring
+- **Persona:** backend-engineer — **REQ:** REQ-4-002
+- **Files:** root `package.json` (update), `turbo.json` (update), `pnpm-workspace.yaml` (verify apps/* already covered — no change expected)
+- **Action:** Add `cli:dev`, `cli:test`, `cli:build`, `cli:typecheck`, `cli:lint` root scripts mirroring the `ai:*` passthrough pattern (D-022/D-037). Turbo tasks for the CLI package: `build` (dependsOn `^build`, outputs `dist/**`), `test`, `typecheck`, `lint` (cache:false, outputs:[] for test — same shape as ai-service). No changes to existing ai:* tasks.
+- **Verify:** `pnpm cli:test` + `pnpm cli:typecheck` green from repo root; `pnpm build` still green for web+ai-service (turbo graph unaffected); `pnpm ai:test` still green.
+
+### Wave 2: Commands (depends on Wave 1)
+
+#### Task 1-2-01: doctor command
+- **Persona:** cli-engineer — **REQ:** REQ-4-001
+- **Files:** `apps/cli/src/commands/doctor.ts`, `apps/cli/tests/doctor.test.ts`
+- **Action:** Checks (each with actionable hint): node ≥18 (`process.version`), pnpm ≥8 on PATH (`pnpm --version`), python3 ≥3.11 (`python3 --version` parse), git (`git --version`), corepack available-or-pnpm-present nuance folded into pnpm check, `unshare` binary on PATH (`which unshare` — sandbox fabric needs it; hint explains what breaks without it). Sequential execution with per-check timeout; summary line; exit 1 if any ✗. Runs from any cwd (no repo required — pure environment check).
+- **Verify:** unit tests with injected spawn results: all-pass → exit 0 + summary; missing pnpm → ✗ + hint + exit 1; missing unshare → ✗ with sandbox-specific hint.
+
+#### Task 1-2-02: bootstrap command
+- **Persona:** cli-engineer — **REQ:** REQ-4-002
+- **Files:** `apps/cli/src/commands/bootstrap.ts`, `apps/cli/src/lib/spawn.ts`, `apps/cli/tests/bootstrap.test.ts`
+- **Action:** `spawn.ts`: `run(cmd, args, {timeoutMs, cwd, env})` — promisified child_process.spawn, inherited stdio, timeout kill (SIGTERM→SIGKILL escalation), returns `{code}`; throws never (codes always returned). `bootstrap.ts` steps (each logged before/after): (1) locate repo root (walk up for pnpm-workspace.yaml; error with hint if not in a clone); (2) `pnpm install` at root; (3) delegate ai-service venv to `apps/ai-service/scripts/bootstrap.sh` via spawn with generous timeout (10 min) — **zero pip/venv logic in the CLI** (A-202); (4) copy `.env.example` → `.env` if absent (preserve existing; report created vs kept); (5) validate optional keys in `.env` vs template — warn-only (A-210); never touch `.ciagent/.env.secrets`; (6) print next-steps (`nextcraft verify`, `nextcraft dev`). Idempotent: every step safe to re-run.
+- **Verify:** unit tests with stub spawn: step order, env copy semantics (absent → create, present → keep), timeout path returns failure code, secrets file never written; `pnpm cli:test` green.
+
+#### Task 1-2-03: verify + dev commands
+- **Persona:** cli-engineer — **REQ:** REQ-4-002
+- **Files:** `apps/cli/src/commands/verify.ts`, `apps/cli/src/commands/dev.ts`, `apps/cli/tests/verify.test.ts`
+- **Action:** `verify.ts` health checks (each runnable + reported): ai-service venv python imports (`import ai_service` via venv python), uvicorn present in venv, ports 3000/8420 free (net stat via node), `.env` exists with AI_PORT parseable, `pnpm build` dry readiness (turbo graph parses — run `turbo build --dry=json` cheap check or typecheck-only default; choose the cheap one). Summary + exit code. `dev.ts`: locate repo root, exec passthrough to `apps/ai-service/scripts/dev.sh` with **inherited stdio and signals** (Ctrl+C semantics), no timeout (long-running); document that web dev server runs via `pnpm dev` separately (dev.sh owns ai-service only).
+- **Verify:** unit tests: verify aggregates check results → exit codes; dev spawns dev.sh with signal passthrough assertions (mock spawn).
+
+### Wave 3: Integration review (depends on Wave 2)
+
+#### Task 1-3-01: CLI security + integration review pass
+- **Persona:** security-auditor — **REQ:** REQ-4-001, REQ-4-002
+- **Files:** `apps/cli/src/lib/spawn.ts` (review; patch if defect), `apps/cli/src/commands/bootstrap.ts` (review), `apps/cli/tests/**` (add regression if defect found)
+- **Action:** STRIDE pass on the CLI surface: spawn injection (args never through shell string — array form only), timeout enforcement, secrets never logged, env template copy doesn't overwrite user edits, no shell=true anywhere, PATH resolution honest errors. Findings → P0 patches now with regression tests; P1+ noted for final-phase review.
+- **Verify:** `pnpm cli:test` green incl. any added regressions; `grep -rn "shell: *true" apps/cli/src` returns nothing.
+
+### Must-Haves (Phase 1)
+- [ ] `pnpm --filter @nextcraft/cli test` green; `pnpm typecheck` green; `pnpm build` green
+- [ ] doctor: every prerequisite reported with ✓/✗ + actionable hint; exit 1 on any ✗; runs outside a repo clone
+- [ ] bootstrap: composes scripts/bootstrap.sh (no pip/venv logic in CLI); idempotent; .env created from template only when absent; optional-key warnings, never blocks; never writes secrets
+- [ ] verify: venv import + uvicorn + ports + env checks with single-glance summary and named failing commands
+- [ ] dev: passthrough with signal inheritance (Ctrl+C stops the stack)
+- [ ] Exit-code contract: 0/1/2; --help everywhere; unknown command → 2
+- [ ] No runtime npm dependencies in apps/cli (dev deps only)
+- [ ] Root `cli:*` scripts work from repo root; ai:* scripts unaffected
+
+---
+
+## Phase 2: Binary Build + Release Pipeline
+
+**Requirements:** REQ-4-003, REQ-4-004
+**Goal:** `nextcraft-linux-x64` SEA binary + sha256 sidecar built reproducibly from the CLI package; one-liner `install.sh` verified end-to-end against a real release; release-asset upload wired so **every ship from now on carries binaries**.
+
+### Wave 1: Binary build (parallel)
+
+#### Task 2-1-01: SEA binary build script (G-101: live-build probe FIRST — mechanism must be proven before the pipeline depends on it)
+- **Persona:** cli-engineer — **REQ:** REQ-4-004
+- **Files:** `apps/cli/scripts/build-binary.mjs`, `apps/cli/package.json` (add `build:binary` script), `apps/cli/.sea-config.json` (or generated in-script)
+- **Action:** **First action of this task: build one real SEA binary end-to-end and run it** (`--version` + `doctor` smoke) before writing the polished script.** Pipeline: esbuild bundle `src/index.ts` → `dist/bundle.cjs` (platform node, target node18, banner shebang, SEA config: `{main: "dist/bundle.cjs", output: "dist/sea-prep.blob", disableExperimentalSEAWarning: true}`) → `node --experimental-sea-config` → copy system node binary → inject blob (`npx postject` with sentinel `NODE_SEA_BLOB_FUSE` fuse, or `dd` fallback) → chmod +x → `dist/nextcraft-linux-x64` → **stamp version from the shipping tag argument** (`NEXTCRAFT_VERSION` injected via esbuild `define`, G-102 — `--version` prints it; absent arg → dev stamp `0.0.0-dev`) → `shasum -a 256` → `dist/nextcraft-linux-x64.sha256`. Fallback (documented, scripted, honest): if SEA injection fails, python3 zipapp builds `nextcraft-linux-x64.pyz` (requires python3 on target — install.sh handles both asset shapes and the docs say so; NO silent claim of node-less operation, G-101).
+- **Verify:** `pnpm --filter @nextcraft/cli build:binary` produces the binary; `./dist/nextcraft-linux-x64 --version` runs **with node absent from PATH** (test via `env -i /bin/sh -c 'PATH=/usr/bin:/bin ...'` sandbox or by temporarily stripping PATH in a subprocess test); sha256 file matches `shasum -c`.
+
+#### Task 2-1-02: Release-asset upload helper
+- **Persona:** cli-engineer — **REQ:** REQ-4-004
+- **Files:** `scripts/release-assets.sh`, `apps/cli/tests/release-assets.test.ts` (fixture-level)
+- **Action:** Given a tag: build binary (Task 2-1-01), resolve GITEA_TOKEN from `.env`/`.env.secrets`/`.env.*` **via the secrets loader only** (never shell env — v1.8 root cause), create/locate the Gitea release via API, upload both assets (`POST /api/v1/repos/{owner}/{repo}/releases/{id}/assets?name=...` multipart). Bounded retry (3) per config.ship.max_release_retries; token never echoed; failure = non-blocking escalation message (release_pending semantics) — tag+merge already complete the ship.
+- **Verify:** fixture test: token resolution order (.env.secrets wins over .env; shell env NEVER consulted — assert with a poisoned env var fixture); dry-run mode prints the exact curl-multipart it would send (no net in tests).
+
+#### Task 2-1-03: install.sh one-liner
+- **Persona:** cli-engineer — **REQ:** REQ-4-003
+- **Files:** `scripts/install.sh`, `apps/cli/tests/install-script.test.ts`
+- **Action:** POSIX sh (no bashisms — dash-safe): `set -eu`; platform check (uname linux + x86_64; else print source-bootstrap path + exit 0 — a graceful no-op, not an error); resolve latest release via Gitea API (`curl -fsSL .../releases/latest`, parse `tag_name` + asset `browser_download_url`s with sed/grep — no jq dependency); **match assets by EXACT name** (`nextcraft-linux-x64`, `nextcraft-linux-x64.sha256` — any parse/lookup miss = degrade to source-bootstrap instructions, exit 0, G-103 — never a name-approximate install); handle the zipapp asset shape (`nextcraft-linux-x64.pyz` + sidecar) when the binary is absent, printing the python3 requirement honestly; download both assets to `mktemp -d` (trap cleanup EXIT); **verify sha256 before anything else** (`shasum -a 256 -c` or sha256sum); on mismatch → hard stop, explicit "do not run" message, exit 1; install to `~/.local/bin` (mkdir -p; `--dest` override); PATH hint when missing (print exact export line); print the binary's own `--version` output (G-102: must equal the resolved release tag — mismatch = install-time integrity stop) + `nextcraft doctor` next-step. No-binary-asset path: print the git-clone + scripts/bootstrap.sh instructions + exit 0. Zero secrets required (public release assets).
+- **Verify:** unit tests over the script's pure helpers extracted where feasible; **live E2E in Task 2-3-01**. `sh -n scripts/install.sh` syntax-clean; `dash scripts/install.sh --help` safe if dash present.
+
+### Wave 2: Ship-flow integration (depends on Wave 1)
+
+#### Task 2-2-01: Wire binaries into every ship (G-104: enforcement, not prose)
+- **Persona:** backend-engineer — **REQ:** REQ-4-004
+- **Files:** `.ciagent/config.json` (no schema change needed — release section already configured), this repo's ship procedure notes (update `.ciagent/ARCHITECTURE.md` Build Order note if needed), `scripts/release-assets.sh` (finalize from 2-1-02)
+- **Action:** Establish the ship-time contract going forward: after every phase ship (tag + merge complete = ship gate per config.ship), run `scripts/release-assets.sh <tag>` to attach binary + checksum to the freshly created release. **G-104:** this run is MANDATORY-ATTEMPTED on every release from v0.3.2 onward — best-effort/non-blocking like release creation (release_pending escalation on exhaustion), logged in the ship commit, and the P4 final audit gate includes "milestone release carries both assets" as an explicit check. This makes "ongoing binaries" a property of the pipeline, not a one-off.
+- **Verify:** The P2 ship itself executes the step against tag v0.3.2 (live validation — see Ship).
+
+### Wave 3: End-to-end validation (depends on Wave 2)
+
+#### Task 2-3-01: Install E2E against the live release
+- **Persona:** security-auditor — **REQ:** REQ-4-003
+- **Files:** `apps/cli/tests/install-e2e.test.ts` (marked slow/e2e), `apps/cli/README.md` (install internals section)
+- **Action:** Live E2E after the v0.3.2 release exists (run post-ship, documented as the verify gate for this phase's asset path): fresh HOME tmpdir → run install.sh → assert binary at `$HOME/.local/bin/nextcraft`, `--version` output equals the release tag (G-102 integrity assertion), checksum verified path taken (tamper test: flip a byte in a local fixture download → script refuses + exits 1). Record the transcript in the phase verify commit. If the live release isn't reachable at verify time, run the full local equivalent (serve assets from a fixture dir via `python3 -m http.server` + FORGE_BASE override) and mark live re-check as a P1 follow-up.
+- **Verify:** E2E green locally (fixture server path mandatory in tests — no test depends on the live forge); tamper-rejection proven; transcript recorded.
+
+### Must-Haves (Phase 2)
+- [ ] **G-101:** a real SEA binary built + smoke-run BEFORE the pipeline depends on it; if SEA fails, zipapp is primary and docs state the python3 requirement
+- [ ] **G-102:** binary `--version` reports the shipping tag (stamped at build); install E2E asserts version == release tag
+- [ ] `pnpm --filter @nextcraft/cli build:binary` produces `nextcraft-linux-x64` + `.sha256`; binary runs without node on PATH (`--version`, `doctor` smoke)
+- [ ] **G-103:** `sh -n scripts/install.sh` clean; dash-safe; exact-name asset matching; platform mismatch → graceful source-bootstrap path (exit 0)
+- [ ] Checksum verified before install; tamper → hard stop with explicit warning (E2E-proven)
+- [ ] install.sh resolves latest release + assets from the Gitea API with zero secrets and no jq
+- [ ] release-assets.sh resolves GITEA_TOKEN from .env* files only (never shell env — tested with poisoned env)
+- [ ] **G-104:** v0.3.2 release carries both assets (live validation at ship); upload failure is non-blocking escalation, attempted + logged every release
+- [ ] `pnpm build`, `pnpm typecheck`, `pnpm cli:test` all green
+
+---
+
+## Phase 3: Install Docs + Fresh-Clone E2E
+
+**Requirements:** REQ-4-005
+**Goal:** README quickstart + CLI reference matching the tested reality exactly, plus a fresh-clone E2E test proving the happy path end-to-end.
+
+### Wave 1: Fresh-clone E2E (drives doc accuracy)
+
+#### Task 3-1-01: Fresh-clone bootstrap E2E
+- **Persona:** cli-engineer — **REQ:** REQ-4-005
+- **Files:** `apps/cli/tests/fresh-clone-e2e.test.ts` (slow/e2e-marked)
+- **Action:** In a `mktemp -d` sandbox: `git clone` the repo locally (file:// clone of HEAD — no network), run `pnpm --filter @nextcraft/cli dev -- doctor` (or the built binary from P2) → then `bootstrap` → then `verify`, asserting each step's exit codes and key output markers. Skips gracefully when network-dependent steps are unavailable (CI marker). Documents the exact happy path the README will state.
+- **Verify:** E2E green locally (clone of the working tree); output transcript matches README claims (cross-checked in 3-2-01).
+
+### Wave 2: Documentation (depends on Wave 1 transcript)
+
+#### Task 3-2-01: README quickstart + CLI reference
+- **Persona:** backend-engineer — **REQ:** REQ-4-005
+- **Files:** root `README.md` (update quickstart section), `apps/cli/README.md` (CLI reference)
+- **Action:** Root README quickstart: the one-liner (exact tested URL), then doctor → bootstrap → verify → dev sequence with expected outputs; source-bootstrap alternative documented (clone + scripts). apps/cli README: every command, flags, exit codes, the env-template copy semantics, optional-key warning semantics, secrets policy (never generated/committed; .ciagent/.env.secrets location), binary install internals, troubleshooting table keyed to actual failure modes observed in E2E.
+- **Verify:** Every command line in both READMEs is copy-paste runnable — verified against the 3-1-01 transcript; doc drift check: no references to commands/flags that don't exist in `--help` output.
+
+### Must-Haves (Phase 3)
+- [ ] Fresh-clone E2E green: doctor → bootstrap → verify sequence from a clean clone
+- [ ] README quickstart matches the E2E transcript exactly (no aspirational docs)
+- [ ] CLI reference covers all 4 commands + --help/--version + exit codes
+- [ ] `pnpm build`, `pnpm typecheck`, `pnpm test` (all suites) green
+
+---
+
+## Phase 4: Final Review + Ship (milestone release v0.3.4)
+
+1. Branch gate → `phase/04-final-review-ship`.
+2. Multi-persona review across the milestone (correctness, testing, security, performance, maintainability, adversarial) — P0 auto-fixed, P1+ fixed in this phase.
+3. Audit: reconstruction test (.ciagent files ↔ git log), file discipline, branch hygiene, commit discipline, P0-review flags resolved, **G-104 gate: milestone release v0.3.4 carries `nextcraft-linux-x64` + `.sha256` assets**.
+4. Milestone ship: merge phase/04 → milestone/v0.4-distribution; merge milestone → main; tag **v0.3.4** (= milestone release); attach binary + checksum assets (the ongoing-binaries contract); release notes with full milestone summary (all phases, all REQ-4-001..005, the "ongoing binaries from now on" statement, v0.5 deferral list per D-016); delete all milestone/phase branches.
+5. Complete: REQUIREMENTS.md REQ-4-001..005 → complete; ROADMAP.md v0.4 → complete; checkpoint cleared.
+
+## Must-Haves (Milestone)
+- [ ] One-liner installs a working binary from the live Gitea release (E2E-proven, tamper-tested)
+- [ ] Fresh clone → doctor → bootstrap → verify → dev: the full happy path green from a clean environment
+- [ ] Every release from v0.3.2 onward carries `nextcraft-linux-x64` + `.sha256` assets
+- [ ] Zero runtime npm deps in the CLI; secrets only ever from .env* files; never in code/logs/commits
+- [ ] All suites green: `pnpm build`, `pnpm typecheck`, `pnpm ai:test`, `pnpm cli:test`

.ciagent/PROJECT.md

@@ -8,45 +8,64 @@ Nextcraft is an AI-native outcome school where graduates prove what they can bui
 
 ---
 
-## Current Milestone: v0.3 — Credential Engines
+## Current Milestone: v0.5 — Real Server Voice, KYC/Identity, Sandbox Environments, Seq-Lease
 
-**Scope:** Replace v0.2's mock engine inputs with real credential engines. Build the sandbox fabric (sandboxed IDE / design tool / simulation), the live in-environment build-telemetry pipeline, the process-trace grading engine, per-learner variant task generation, and the oral/voice defense with AI examiner. Lab/Assessor/Proctor agents move from mock inputs to real engine inputs; the six tutor agents operate on authentic telemetry and artifacts.
+**Scope (deferred seams from D-016, to be specified at v0.5 Phase 0):** real server STT/TTS (`openai-audio` voice provider, CUT-1/G-7 seam), KYC/identity verification + age-gating backend (REQ-F-017), design/simulation sandbox environments (REQ-F-021 remainder), exec-telemetry seq-lease/replay-margin fix.
+
+## Prior Milestone: v0.4 — Distribution & Bootstrap CLI (COMPLETE, shipped as v0.3.4)
+
+**Scope (founder directive, 2026-09-12):** Streamline installing Nextcraft. Ship a bootstrap CLI with a single-liner install script, and publish release binaries on an ongoing basis for every release going forward.
+
+**Delivered:** `nextcraft` CLI — `doctor` (prerequisite checks), `bootstrap` (deps + venv + env from templates + key validation), `verify` (health check), `dev` (thin passthrough to scripts/dev.sh); one-liner install script downloading the linux x64 binary from the latest Gitea release with sha256 + version integrity gates; binary release pipeline attached to every ship from v0.3.2 onward; install/quickstart documentation backed by a fresh-clone E2E test. All 5 requirements (REQ-4-001..005) complete.
+
+**Status of v0.3:** Complete and shipped (v0.2.8). Credential engines live: namespace-isolated sandbox fabric, live build telemetry, process-trace grading, seeded variants, oral defense; real learner build/defense/grading surfaces.
 
 **Status of v0.2:** Complete and shipped (v0.2.0). Six AI tutor agents live over mockengine inputs (D-015).
 
-**Deferred from earlier plan:** REQ-F-017 (identity verification + 16+/18+ age-gating) is explicitly deferred to a later milestone per founder directive. Age-gating remains the v0.1-style visual flow mockup; no real KYC backend is built in v0.3.
+**Deferred from earlier plan:** REQ-F-017 (identity verification + age-gating KYC), real server STT/TTS, design/simulation sandbox environments, and the exec-telemetry seq-lease are all deferred to v0.5. Age-gating remains the v0.1-style visual flow mockup.
 
 **Tech stack:** v0.1 TS monorepo (pnpm/turborepo, Next.js) + v0.2 Python FastAPI ai-service + new credential-engine services (sandbox fabric orchestrator, telemetry ingest, grading engine) in Python/TypeScript as determined at RESEARCH.
 
 ---
 
-## Requirements (Validated)
+## v0.4 Requirements (Complete)
 
-The following requirements have been validated during specification and are locked for milestone v0.3 (REQ-F-007..010 and REQ-F-021 activated from the deferred pool; REQ-F-017 deferred per founder directive):
+All 5 v0.4 requirements (REQ-4-001..005) are complete and shipped as v0.3.4:
+1. Bootstrap CLI — `nextcraft` executable with `doctor` / `bootstrap` / `verify` / `dev` (REQ-4-001, REQ-4-002)
+2. One-liner install — `curl | sh` fetching the linux x64 binary from the latest Gitea release with sha256 + version integrity verification (REQ-4-003)
+3. Ongoing release binaries — every release from v0.3.2 onward ships the CLI binary + checksum as release assets (REQ-4-004)
+4. Install documentation — README quickstart + CLI reference verified by a fresh-clone E2E test (REQ-4-005)
 
-1. Sandbox fabric — sandboxed IDE, design tool, and simulation environments with isolated execution and lifecycle management (REQ-F-021)
-2. Live build telemetry — in-environment capture of process events (keystrokes, commands, file diffs, run/test results) streamed to ai-service (REQ-F-010)
-3. Process-trace grading engine — grade artifacts from their process traces, not just final output (REQ-F-007); feeds the Assessor agent real inputs
-4. Variant task generation — per-learner task variants so no two learners receive identical prompts (REQ-F-008)
-5. Oral/voice defense — AI examiner conducts spoken defense of submitted work (REQ-F-009); feeds the Proctor/Mentor agents
-6. Agent re-grounding — Lab/Assessor/Proctor consume real engine inputs (telemetry, traces, defenses) instead of v0.2 mocks
-7. Learner surface integration — wire the v0.1 sandbox + assessment mockups to the real engines (build/run in-browser, live telemetry, live defense)
+## v0.3 Requirements (Complete)
 
-## v0.2 Requirements (Complete)
-
-All 12 v0.2 requirements (REQ-2-001..012) are complete and shipped as v0.2.0. See REQUIREMENTS.md traceability matrix.
+All 8 v0.3 requirements (REQ-3-001..008) are complete and shipped as v0.2.8. See REQUIREMENTS.md traceability matrix.
 
 ## v0.1 Requirements (Complete)
 
 All 28 v0.1 requirements (REQ-001..028) are complete and shipped as v0.1.0. See REQUIREMENTS.md traceability matrix.
 
+## Clarified Assumptions (v0.4 CLARIFY stage, full autonomy — auto-resolved)
+
+| # | Ambiguity | Resolution | Confidence |
+|---|-----------|------------|-------------|
+| A-201 | CLI language/toolchain for the binary? | **Probe-driven at RESEARCH** — Go → Rust → Node SEA → Python zipapp fallback chain; spec stays toolchain-agnostic so PLAN locks the probe-verified toolchain | 0.70 |
+| A-202 | Does bootstrap replace scripts/bootstrap.sh? | **No — reuse it.** CLI wraps existing `scripts/bootstrap.sh` + `scripts/dev.sh` via subprocess; zero orchestration logic duplicated in the CLI (thin passthrough pattern) | 0.85 |
+| A-203 | Where does the one-liner fetch the binary? | **Gitea latest-release API** (`/repos/{owner}/{repo}/releases/latest`) → download `nextcraft-linux-x64` + `.sha256` asset; repo raw serves `install.sh` as the stable URL | 0.80 |
+| A-204 | Install target + PATH? | **~/.local/bin** (XDG-style, no sudo), PATH hint printed when missing; `--dest` override flag | 0.85 |
+| A-205 | Binary "ongoing releases" scope? | **Every ship from v0.4 onward** attaches `nextcraft-linux-x64` + sha256 sidecar to the Gitea release — the ship workflow gains an asset step; retroactive binaries for old releases NOT required | 0.90 |
+| A-206 | No binary available yet / non-linux? | **Graceful degradation**: install script prints source-bootstrap instructions (git clone + scripts/bootstrap.sh) — never a hard fail | 0.88 |
+| A-207 | Checksum trust root? | **sha256 sidecar shipped as a release asset next to the binary** (same release, same channel); script verifies download against it. Signature/PKI out of scope for v0.4 (single forge, TLS transport) | 0.75 |
+| A-208 | Which prerequisites does doctor check? | node ≥18, pnpm ≥8, python3 ≥3.11, git, `unshare` availability (sandbox fabric needs it) — versions from the existing bootstrap tooling, not invented | 0.85 |
+| A-209 | Does `dev` manage multiple processes? | **No.** Thin passthrough to scripts/dev.sh only — the CLI stays bootstrap-scoped (D-016); orchestration remains in dev.sh | 0.82 |
+| A-210 | `.env.secrets` handling by bootstrap? | **Template copy only for `.env.example` → `.env`; secrets NEVER generated, NEVER committed; bootstrap validates presence of optional keys and warns (not blocks) when missing — mock-first providers keep the stack runnable** | 0.90 |
+
 ## Clarified Assumptions (v0.3 CLARIFY stage, full autonomy — auto-resolved)
 
 | # | Ambiguity | Resolution | Confidence |
 |---|-----------|------------|-------------|
 | A-101 | Sandbox isolation technology? | **`unshare` user+mount+pid+net namespace subprocess isolation** per sandbox (probe-verified: in-ns uid=0, network fully isolated with 0 interfaces, writes land in an isolated bind-mounted workdir; proc-remount is not permitted in this context but is not required). No Docker/Podman/VMs — none present on the box; no sudo. A `SandboxBackend` protocol keeps a future containerd swap possible. Falls back further to a plain chroot-free subprocess with a cwd-jail if userns ever unavailable (tested path is userns). | 0.8 |
 | A-102 | Sandbox scope in v0.3? | **Coding IDE only** (web terminal + file tree + run/test). The "design tool" and "simulation" environments specified in REQ-F-021 are deferred to v0.4 — a single real build environment is enough to prove the credential pipeline end-to-end (telemetry → trace → grade → defense). | 0.75 |
-| A-103 | Live in-browser build UX? | **WebSocket xterm.js terminal** attached to the bwrap sandbox shell + HTTP file-tree/CRUD + run/test buttons. No full Monaco LSP in v0.3 — a code editor with syntax highlight (existing) + real shell is sufficient and far cheaper. | 0.72 |
+| A-103 | Live in-browser build UX? | **Run/Test buttons executing in the namespace sandbox + HTTP file-tree/CRUD + read-only exec-output panel** (CUT-2/G-8 — the interactive xterm.js shell relay is deferred to v0.4; `@xterm/*` is not a v0.3 dependency). No full Monaco LSP in v0.3 — a code editor with syntax highlight (existing) is sufficient and far cheaper. | 0.72 |
 | A-104 | Telemetry transport? | **WebSocket** from sandbox to a new ingestion endpoint on ai-service for live events; **SQLite-backed** ordered event log (`ai_service/telemetry/`) gives durability + at-least-once delivery + replay. Events carry monotonic `seq` per (learner,task) so gaps are detectable. | 0.8 |
 | A-105 | Where do traces live? | **SQLite** (`ai_service` data dir), introducing the first real persistence. SQLModel/SQLAlchemy for typed access. Chosen over Postgres because solo-founder + single box + low write volume; the `TraceStore` protocol is Postgres-migration-ready like SessionStore was. | 0.75 |
 | A-106 | Process-trace grading model? | **LLM-based grader**: structure the trace into a compact timeline digest (command categories, error/fix cycles, idle gaps, test passes) → Assessor-style rubric prompt → structured score via existing D-020 JSON defense. Deterministic features (test pass/fail, edit count) computed in code, not left to the LLM. | 0.7 |
@@ -54,7 +73,7 @@ All 28 v0.1 requirements (REQ-001..028) are complete and shipped as v0.1.0. See
 | A-108 | Voice defense — STT/TTS providers? | **Provider-agnostic, mock-first like the LLM layer (D-014).** Real path: browser `MediaRecorder` → audio to ai-service → **OpenAI-compatible `/audio/transcriptions`** (Whisper STT) and **`/audio/speech`** (TTS) against ollama-cloud or a compatible endpoint; fallbacks: browser `SpeechRecognition`/`speechSynthesis` when no server keys. `VoiceProvider` protocol + deterministic mock (returns canned transcript) so tests never call a voice API. | 0.62 |
 | A-109 | Defense dialogue shape? | Reuse BaseAgent: an `Examiner` agent (seventh agent) streams examiner questions over the existing SSE pipeline; integrity signals (long pauses, off-scope answers, reading-from-notes cadence) emitted alongside the transcript to Proctor. | 0.8 |
 | A-110 | KYC / age-gating in v0.3? | **Deferred per founder directive.** No real identity backend. Age-gating stays the v0.1 visual flow mockup. Personas omit a security-engineer; security review via verifier + Phase 7 secrets-hygiene checklist. **Abuse control is NOT deferred with KYC (G-5):** v0.3 ships per-learner sandbox caps (`AI_SANDBOX_MAX_PER_LEARNER`), a global create-rate cap, and a server-side `learner_id` allowlist (`AI_LEARNER_ALLOWLIST`) so the unauthenticated surface cannot exhaust shared NPROC/disk. Documented in the release note. | 0.98 |
-| A-111 | New services vs extend ai-service? | **Extend ai-service**, don't fork new Python apps. Telemetry ingestion, trace grading, variant generation, voice, and sandbox orchestration all live as new modules in `apps/ai-service` (they share the LLM provider pool + config + session infra). Only the in-sandbox capture agent is a separate tiny Python process shipped into the bwrap environment. | 0.82 |
+| A-111 | New services vs extend ai-service? | **Extend ai-service**, don't fork new Python apps. Telemetry ingestion, trace grading, variant generation, voice, and sandbox orchestration all live as new modules in `apps/ai-service` (they share the LLM provider pool + config + session infra). Only the in-sandbox capture agent is a separate tiny Python process shipped into the namespace sandbox. | 0.82 |
 | A-112 | Sandbox on a single dev/school box — capacity? | v0.3 targets **1–5 concurrent sandboxes** (founder + pilot learners). No horizontal scaling, no queue. Concurrency guard returns 503 when full. Scaling is post-MVP. | 0.8 |
 
 ## Clarified Assumptions (v0.2 CLARIFY stage, full autonomy — auto-resolved)
@@ -133,6 +152,7 @@ The following remain deferred beyond v0.3 and will be activated in subsequent mi
 | D-013 | v0.1 prototype founder-agreed; D-001 business-logic gate unlocked | Founder approved starting v0.2 with AI Tutor Architecture, which constitutes agreement of the v0.1 prototype per D-001. Recorded at v0.2 SPECIFY. | Business logic authorized from v0.2 onward |
 | D-014 | Provider-agnostic LLM layer; ollama-cloud as initial provider | OpenAI-compatible client abstraction with pluggable providers: ollama-cloud (https://ollama.com/v1, default), local OpenAI-compatible endpoint, deterministic mock (tests/CI). Keys in gitignored .ciagent/.env.secrets, never in code or commits. | apps/ai-service llm package with 3 providers; default=ollama-cloud |
 | D-015 | All six agents implemented as real LLM services; engines mocked | Coach/Tutor/Mentor fully real. Lab/Assessor/Proctor are real LLM logic over mock inputs (simulated telemetry, pre-baked artifacts) since sandbox fabric, assessment engine, and identity verification are v0.3+. Consistent with v0.1's mock-data approach. | REQ-F-001..006 complete in v0.2; real engines deferred to v0.3+ |
+| D-016 | v0.4 = Distribution & Bootstrap CLI (founder directive supersedes previously-named v0.4 seams) | Founder directive 2026-09-12: focus this milestone on streamlining install, a bootstrap CLI with a one-liner install script, ongoing release binaries. Real server STT/TTS, KYC, design/simulation envs, seq-lease move to v0.5. | Milestone scope locked at SPECIFY; binary = CLI-only, linux x64 |
 
 ---
 

.ciagent/REQUIREMENTS.md

@@ -1,5 +1,22 @@
 # Nextcraft — REQUIREMENTS.md
 
+## v0.4 Requirements (Complete — Distribution & Bootstrap CLI, shipped as v0.3.4)
+
+### Bootstrap CLI
+
+| ID | Description | Priority | Phase | Status |
+|----|-------------|----------|-------|--------|
+| REQ-4-001 | `nextcraft` CLI (linux x64 binary): `doctor` command checking prerequisites (node, pnpm, python3, git, unshare) with actionable error messages | critical | 1 | complete |
+| REQ-4-002 | `bootstrap` command: pnpm install, ai-service venv + pinned deps, .env from templates, key validation, .env.secrets handling; `verify` health check (ports, imports, builds); `dev` thin passthrough to scripts/dev.sh | critical | 1 | complete |
+
+### Distribution
+
+| ID | Description | Priority | Phase | Status |
+|----|-------------|----------|-------|--------|
+| REQ-4-003 | One-liner install script (`curl -fsSL <url> \| bash`): detects linux x64, resolves latest release from Gitea API, downloads binary + checksum, verifies sha256, installs to ~/.local/bin (PATH hint), degrades to source-bootstrap instructions when no binary | critical | 2 | complete |
+| REQ-4-004 | Binary release pipeline: reproducible linux x64 build script, sha256 checksum sidecar, upload as release assets on every ship from v0.4 onward (ongoing binaries requirement) | critical | 2 | complete |
+| REQ-4-005 | Install + quickstart documentation: README one-liner quickstart, CLI command reference, fresh-clone-to-running-stack end-to-end verification | high | 3 | complete |
+
 ## v0.3 Requirements (Credential Engines)
 
 ### Sandbox & Telemetry
@@ -8,22 +25,22 @@
 |----|-------------|----------|-------|--------|
 | REQ-3-001 | Sandbox fabric: isolated per-learner execution environments (sandboxed IDE, design tool, simulation) with lifecycle management | critical | 1 | complete |
 | REQ-3-002 | Sandbox isolation + resource limits: per-learner isolation boundary, CPU/memory quotas (rlimits), wall-clock time quota, disk-quota via per-sandbox workdir usage sweep (best-effort, not kernel-enforced), no cross-tenant access, snapshot support. **Known gap (v0.3): per-sandbox pids and hard disk caps are NOT kernel-enforceable without cgroup delegation/sudo — documented as accepted risk** | critical | 1 | complete |
-| REQ-3-003 | Live build telemetry: in-environment capture of process events (commands, file diffs, run/test results, activity) streamed reliably to ai-service with per-learner trace persistence | critical | 2 | pending |
+| REQ-3-003 | Live build telemetry: in-environment capture of process events (commands, file diffs, run/test results, activity) streamed reliably to ai-service with per-learner trace persistence | critical | 2 | complete |
 
 ### Credential Engines
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-3-004 | Process-trace grading engine: grade artifacts from their full process traces; rubric-aligned structured scores; feeds Assessor real inputs | critical | 3 | pending |
-| REQ-3-005 | Variant task generation: per-learner task variants (no two learners get identical prompts); variant seed registry; difficulty normalization | high | 4 | pending |
-| REQ-3-006 | Oral/voice defense: AI examiner conducts spoken defense (STT → dialogue → TTS); transcript + integrity signals captured; feeds Proctor/Mentor | high | 5 | pending |
+| REQ-3-004 | Process-trace grading engine: grade artifacts from their full process traces; rubric-aligned structured scores; feeds Assessor real inputs | critical | 3 | complete |
+| REQ-3-005 | Variant task generation: per-learner task variants (no two learners get identical prompts); variant seed registry; difficulty normalization | high | 4 | complete |
+| REQ-3-006 | Oral/voice defense: AI examiner conducts spoken defense (STT → dialogue → TTS); transcript + integrity signals captured; feeds Proctor/Mentor | high | 5 | complete |
 
 ### Agent Re-grounding & Integration
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-3-007 | Agent re-grounding: Lab consumes live telemetry; Assessor consumes grading-engine output; Proctor consumes telemetry + defense integrity signals (replace v0.2 mocks) | critical | 6 | pending |
-| REQ-3-008 | Learner surface integration: sandbox mockup → real in-browser build/run with live telemetry; assessment mockup → live defense + live grading | critical | 6 | pending |
+| REQ-3-007 | Agent re-grounding: Lab consumes live telemetry; Assessor consumes grading-engine output; Proctor consumes telemetry + defense integrity signals (replace v0.2 mocks) | critical | 6 | complete |
+| REQ-3-008 | Learner surface integration: sandbox mockup → real in-browser build/run with live telemetry; assessment mockup → live defense + live grading | critical | 6 | complete |
 
 ---
 
@@ -65,58 +82,58 @@
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
 | REQ-001 | Monorepo scaffolding: pnpm workspaces, turborepo, Next.js app, TypeScript config, ESLint, Prettier | critical | 1 | complete |
-| REQ-002 | Shared component library: design tokens, Button, Input, Card, Badge, Avatar, Dialog, Navigation, Table, Tabs, Progress, Tooltip, Skeleton, Toast | critical | 1 | pending |
-| REQ-003 | Mock data layer: typed mock data for competency stacks, job listings, candidate profiles, employers, learner progress | critical | 1 | pending |
-| REQ-004 | Routing and navigation: App Router route groups for (learner), (marketplace), (employer), (admin); shared layout components; cross-surface navigation | critical | 1 | pending |
-| REQ-005 | Responsive layout system: mobile, tablet, desktop breakpoints; container components; grid system | high | 1 | pending |
+| REQ-002 | Shared component library: design tokens, Button, Input, Card, Badge, Avatar, Dialog, Navigation, Table, Tabs, Progress, Tooltip, Skeleton, Toast | critical | 1 | complete |
+| REQ-003 | Mock data layer: typed mock data for competency stacks, job listings, candidate profiles, employers, learner progress | critical | 1 | complete |
+| REQ-004 | Routing and navigation: App Router route groups for (learner), (marketplace), (employer), (admin); shared layout components; cross-surface navigation | critical | 1 | complete |
+| REQ-005 | Responsive layout system: mobile, tablet, desktop breakpoints; container components; grid system | high | 1 | complete |
 
 ### Learner Surface
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-006 | Landing page: hero, value proposition, program highlights, how-it-works (Byte→Build→Demonstrate→Defend), testimonials mockup, CTA to program catalog | critical | 2 | pending |
-| REQ-007 | Program catalog: grid of competency stacks (AI Orchestration Engineer, AI Safety & Governance Lead, Human-AI Product Designer, AI-Augmented Field Operator, Computational Sciences Practitioner); stack cards with role descriptions | critical | 2 | pending |
-| REQ-008 | Competency stack view: selected stack detail with 12-18 competencies listed, progress indicators, microcredential badges, mastery status | critical | 2 | pending |
-| REQ-009 | Learner dashboard: active competencies, progress graph, recent artifacts, upcoming defenses, AI tutor chat mockup, milestone tracker | critical | 2 | pending |
-| REQ-010 | Byte tutorial viewer: 3-7 minute micro-tutorial layout with concept panel, worked example panel, code/design/simulation viewer mockup | high | 2 | pending |
-| REQ-011 | Build sandbox mockup: sandboxed IDE/design tool/simulation UI mockup with toolbar, file explorer, editor area, telemetry sidebar (process capture indicators) | high | 2 | pending |
-| REQ-012 | Assessment/defense mockup: rubric display, AI reviewer panel, oral defense interface (voice/mic mockup), process trace timeline, artifact viewer | high | 2 | pending |
+| REQ-006 | Landing page: hero, value proposition, program highlights, how-it-works (Byte→Build→Demonstrate→Defend), testimonials mockup, CTA to program catalog | critical | 2 | complete |
+| REQ-007 | Program catalog: grid of competency stacks (AI Orchestration Engineer, AI Safety & Governance Lead, Human-AI Product Designer, AI-Augmented Field Operator, Computational Sciences Practitioner); stack cards with role descriptions | critical | 2 | complete |
+| REQ-008 | Competency stack view: selected stack detail with 12-18 competencies listed, progress indicators, microcredential badges, mastery status | critical | 2 | complete |
+| REQ-009 | Learner dashboard: active competencies, progress graph, recent artifacts, upcoming defenses, AI tutor chat mockup, milestone tracker | critical | 2 | complete |
+| REQ-010 | Byte tutorial viewer: 3-7 minute micro-tutorial layout with concept panel, worked example panel, code/design/simulation viewer mockup | high | 2 | complete |
+| REQ-011 | Build sandbox mockup: sandboxed IDE/design tool/simulation UI mockup with toolbar, file explorer, editor area, telemetry sidebar (process capture indicators) | high | 2 | complete |
+| REQ-012 | Assessment/defense mockup: rubric display, AI reviewer panel, oral defense interface (voice/mic mockup), process trace timeline, artifact viewer | high | 2 | complete |
 
 ### Marketplace Surface
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-013 | Job board listing: searchable grid of AI-era job listings, filter sidebar (skills, seniority, location, salary), result cards with match score | critical | 3 | pending |
-| REQ-014 | Job detail page: full job description, required competencies, employer info, application CTA, related jobs, AI-matched skills breakdown | critical | 3 | pending |
-| REQ-015 | Employer profile: company overview, logo, description, social links, open positions, company culture mockup | high | 3 | pending |
-| REQ-016 | Search/filter UI: semantic search bar, skill tags, category filters, seniority filter, remote/on-site toggle, saved searches mockup | critical | 3 | pending |
-| REQ-017 | Pricing page: job posting packages (single, bundle, enterprise), talent access plans, subscription tiers, feature comparison table | high | 3 | pending |
+| REQ-013 | Job board listing: searchable grid of AI-era job listings, filter sidebar (skills, seniority, location, salary), result cards with match score | critical | 3 | complete |
+| REQ-014 | Job detail page: full job description, required competencies, employer info, application CTA, related jobs, AI-matched skills breakdown | critical | 3 | complete |
+| REQ-015 | Employer profile: company overview, logo, description, social links, open positions, company culture mockup | high | 3 | complete |
+| REQ-016 | Search/filter UI: semantic search bar, skill tags, category filters, seniority filter, remote/on-site toggle, saved searches mockup | critical | 3 | complete |
+| REQ-017 | Pricing page: job posting packages (single, bundle, enterprise), talent access plans, subscription tiers, feature comparison table | high | 3 | complete |
 
 ### Employer Dashboard
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-018 | Employer dashboard overview: active postings, applicant pipeline, talent matches, analytics mockup (charts, placement stats) | critical | 4 | pending |
-| REQ-019 | Talent search: searchable candidate database with AI-matched filters, candidate cards showing competency stack, microcredentials, artifact count, defense score | critical | 4 | pending |
-| REQ-020 | Candidate profile view: full candidate profile with artifact gallery, process trace summary, oral defense transcripts, competency graph, microcredential verification | high | 4 | pending |
-| REQ-021 | Posting management: create/edit/delete job postings, posting status tracking, applicant list per posting, interview pipeline mockup | high | 4 | pending |
+| REQ-018 | Employer dashboard overview: active postings, applicant pipeline, talent matches, analytics mockup (charts, placement stats) | critical | 4 | complete |
+| REQ-019 | Talent search: searchable candidate database with AI-matched filters, candidate cards showing competency stack, microcredentials, artifact count, defense score | critical | 4 | complete |
+| REQ-020 | Candidate profile view: full candidate profile with artifact gallery, process trace summary, oral defense transcripts, competency graph, microcredential verification | high | 4 | complete |
+| REQ-021 | Posting management: create/edit/delete job postings, posting status tracking, applicant list per posting, interview pipeline mockup | high | 4 | complete |
 
 ### Admin Surface
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-022 | Admin overview: platform metrics (learners, employers, placements, completion rate, NPS), recent activity feed, system health mockup | critical | 5 | pending |
-| REQ-023 | Learner management: searchable learner table, learner detail view, progress tracking, competency completion status, credential issuance log | high | 5 | pending |
-| REQ-024 | Competency graph viewer: interactive visualization of competency stacks and their relationships, node/edge graph using react-flow, stack details on node click | high | 5 | pending |
-| REQ-025 | Marketplace moderation: job posting review queue, employer verification queue, flagged content, content moderation tools mockup | high | 5 | pending |
+| REQ-022 | Admin overview: platform metrics (learners, employers, placements, completion rate, NPS), recent activity feed, system health mockup | critical | 5 | complete |
+| REQ-023 | Learner management: searchable learner table, learner detail view, progress tracking, competency completion status, credential issuance log | high | 5 | complete |
+| REQ-024 | Competency graph viewer: interactive visualization of competency stacks and their relationships, node/edge graph using react-flow, stack details on node click | high | 5 | complete |
+| REQ-025 | Marketplace moderation: job posting review queue, employer verification queue, flagged content, content moderation tools mockup | high | 5 | complete |
 
 ### Polish & Integration
 
 | ID | Description | Priority | Phase | Status |
 |----|-------------|----------|-------|--------|
-| REQ-026 | Cross-surface navigation: role switcher (learner/employer/admin), breadcrumbs, consistent header/footer across all surfaces | critical | 6 | pending |
-| REQ-027 | Visual consistency audit: typography scale, color palette, spacing system, dark mode toggle, accessibility baseline (WCAG AA contrast) | high | 6 | pending |
-| REQ-028 | Component library finalization: Storybook setup, component documentation, prop tables, usage examples | medium | 6 | pending |
+| REQ-026 | Cross-surface navigation: role switcher (learner/employer/admin), breadcrumbs, consistent header/footer across all surfaces | critical | 6 | complete |
+| REQ-027 | Visual consistency audit: typography scale, color palette, spacing system, dark mode toggle, accessibility baseline (WCAG AA contrast) | high | 6 | complete |
+| REQ-028 | Component library finalization: Storybook setup, component documentation, prop tables, usage examples | medium | 6 | complete |
 
 ---
 
@@ -170,18 +187,28 @@
 
 ## Traceability Matrix
 
-### v0.3 (current milestone)
+### v0.4 (current milestone)
+
+| Requirement | Phase | Status |
+|-------------|-------|--------|
+| REQ-4-001 | 1 | complete |
+| REQ-4-002 | 1 | complete |
+| REQ-4-003 | 2 | complete |
+| REQ-4-004 | 2 | complete |
+| REQ-4-005 | 3 | complete |
+
+### v0.3 (complete)
 
 | Requirement | Phase | Status |
 |-------------|-------|--------|
 | REQ-3-001 | 1 | complete |
 | REQ-3-002 | 1 | complete |
-| REQ-3-003 | 2 | pending |
-| REQ-3-004 | 3 | pending |
-| REQ-3-005 | 4 | pending |
-| REQ-3-006 | 5 | pending |
-| REQ-3-007 | 6 | pending |
-| REQ-3-008 | 6 | pending |
+| REQ-3-003 | 2 | complete |
+| REQ-3-004 | 3 | complete |
+| REQ-3-005 | 4 | complete |
+| REQ-3-006 | 5 | complete |
+| REQ-3-007 | 6 | complete |
+| REQ-3-008 | 6 | complete |
 
 ### v0.2 (complete)
 

.ciagent/ROADMAP.md

@@ -2,15 +2,17 @@
 
 ## Overview
 
-**Milestone v0.3** — Credential Engines: Replace v0.2's mock engine inputs with real credential engines. Build the sandbox fabric (sandboxed IDE / design tool / simulation), the live in-environment build-telemetry pipeline, the process-trace grading engine, per-learner variant task generation, and the oral/voice defense with AI examiner. Lab/Assessor/Proctor agents move from mock inputs to real engine inputs.
+**Milestone v0.4 — COMPLETE (shipped as v0.3.4, 2026-09-13).** Distribution & Bootstrap CLI: `nextcraft` CLI (doctor/bootstrap/verify/dev) shipped as a linux x64 SEA binary, one-liner install script with checksum + version integrity gates, and binaries published on **every ongoing release** (v0.3.2 onward). Next milestone: v0.5 (real server STT/TTS + KYC/identity + design/simulation sandbox environments + exec-telemetry seq-lease — the seams deferred out of v0.4 by founder directive D-016).
 
-**Deferred per founder directive:** REQ-F-017 identity verification + age-gating (real KYC backend) is deferred beyond v0.3. Age-gating remains the v0.1 visual flow mockup.
+**Milestone v0.3** — Credential Engines: complete, shipped as v0.2.8 (2026-09-12). Real sandbox fabric, live build telemetry, process-trace grading, per-learner variants, oral defense, real learner surfaces.
+
+**Deferred per founder directive (D-016):** REQ-F-017 identity verification + age-gating (real KYC backend), real server STT/TTS, design/simulation sandbox environments, and the exec-telemetry seq-lease are deferred to v0.5. Age-gating remains the v0.1 visual flow mockup.
 
 **Prior milestone:** v0.2 (ai-tutor-architecture) — complete, shipped as v0.2.0, six tutor agents live over mock engine inputs (D-015).
 
-**Milestone type:** Feature (new credential-engine services + real agent inputs)
-**Tag line:** v0.2.x (patches on the v0.2 line; milestone release as the final v0.2.x patch)
-**Branch:** milestone/v0.3-credential-engines
+**Milestone type:** Feature (new CLI + distribution pipeline)
+**Tag line:** v0.3.x (patches on the v0.3 line; milestone release as the final v0.3.x patch)
+**Branch:** milestone/v0.4-distribution
 
 ---
 
@@ -18,14 +20,11 @@
 
 | # | Name | Status | Depends On | Requirements | Success Criteria |
 |---|------|--------|------------|--------------|------------------|
-| 0 | Pre-execution | in-progress | — | — | Specification, clarify, research, plan complete; .ciagent/ files updated for v0.3 |
-| 1 | Sandbox fabric | complete | 0 | REQ-3-001, REQ-3-002 | Isolated per-learner sandbox environments provisioned (IDE / design / simulation); lifecycle API (create/destroy/snapshot); resource limits enforced; no cross-tenant access |
-| 2 | Live build telemetry | pending | 1 | REQ-3-003 | In-environment capture of process events (commands, file diffs, run/test results, keystroke-level activity) streamed to ai-service; reliable transport; per-learner trace persistence |
-| 3 | Process-trace grading engine | pending | 2 | REQ-3-004 | Grades artifacts from their full process traces (not just final output); emits structured rubric-aligned scores; feeds Assessor real inputs |
-| 4 | Variant task generation | pending | 1 | REQ-3-005 | Per-learner task variants generated so no two learners receive identical prompts; variant seed recorded for grading fairness |
-| 5 | Oral / voice defense | pending | 3 | REQ-3-006 | AI examiner conducts spoken defense of submitted work; STT → dialogue → TTS; transcript + integrity signals captured; feeds Proctor/Mentor |
-| 6 | Agent re-grounding + learner surface integration | pending | 2,3,4,5 | REQ-3-007, REQ-3-008 | Lab/Assessor/Proctor consume real engine inputs; v0.1 sandbox + assessment mockups wired to real engines (in-browser build/run, live telemetry, live defense) |
-| 7 | Final review + ship | pending | 6 | — | Code review clean; audit passes; milestone tagged (v0.2.x final patch); release created on Gitea |
+| 0 | Pre-execution | complete | — | — | Specification, clarify, research, plan complete; .ciagent/ files updated for v0.4 |
+| 1 | Bootstrap CLI core | complete | 0 | REQ-4-001, REQ-4-002 | `nextcraft doctor/bootstrap/verify/dev` work against a fresh clone; unit tests green |
+| 2 | Binary build + release pipeline | complete | 1 | REQ-4-003, REQ-4-004 | Reproducible linux x64 binary + sha256 checksum; one-liner install script; assets uploaded to the Gitea release |
+| 3 | Install docs + fresh-clone E2E | complete | 2 | REQ-4-005 | README quickstart verified end-to-end from a clean environment; fresh clone reaches running stack |
+| 4 | Final review + ship | complete | 3 | — | Code review clean; audit passes; milestone tagged (v0.3.x final patch); release with binary assets created on Gitea |
 
 ---
 
@@ -33,148 +32,107 @@
 
 ### Phase 0: Pre-execution
 
-**Goal:** Establish v0.3 specification, clarify ambiguities, research credential-engine architecture (sandbox isolation, telemetry transport, trace grading, variant generation, voice IO), create detailed plans.
+**Goal:** Establish v0.4 specification (founder directive D-016), clarify ambiguities, research the binary toolchain + Gitea release-asset API + existing bootstrap scripts, create detailed plans, grill adversarially.
 
 **Stages:** SPECIFY → CLARIFY → RESEARCH → PLAN → GRILL → MVP/UX CHECK → SHIP
 
 **Deliverables:**
 - Updated .ciagent/config.json, PROJECT.md, REQUIREMENTS.md, ROADMAP.md, ARCHITECTURE.md, PERSONAS.md, PLAN.md
 
-**Success criteria:** All .ciagent/ files updated for v0.3; phase 0 shipped as first v0.2.x patch.
+**Success criteria:** All .ciagent/ files updated for v0.4; phase 0 shipped as v0.3.0.
 
 ---
 
-### Phase 1: Sandbox Fabric
+### Phase 1: Bootstrap CLI Core
 
-**Goal:** Provision and manage isolated per-learner execution environments.
+**Goal:** A working `nextcraft` CLI with doctor/bootstrap/verify/dev commands, unit-tested against the real monorepo.
 
-**Requirements:** REQ-3-001, REQ-3-002
+**Requirements:** REQ-4-001, REQ-4-002
 
 **Key deliverables:**
-- Sandbox orchestrator service: create/list/destroy/snapshot sandbox instances (IDE, design tool, simulation)
-- Isolation boundary: per-learner containerization or VM-grade isolation; no cross-tenant filesystem/network access
-- Resource limits: CPU/memory/disk/time quotas per sandbox
-- Sandbox lifecycle API consumed by ai-service and the web learner surface
+- `apps/cli` package: `nextcraft` executable (source-runnable in dev, binary-built in P2)
+- `doctor`: checks node ≥18, pnpm, python3 ≥3.11, git, unshare availability — actionable errors, exit codes
+- `bootstrap`: idempotent — pnpm install, ai-service venv + pinned deps (reuses scripts/bootstrap.sh logic), .env from .env.example templates, key validation (warnings not blockers for optional keys), .env.secrets handling
+- `verify`: health check — venv imports, pnpm build readiness, ports free, env vars present
+- `dev`: thin passthrough to scripts/dev.sh (no orchestration logic duplicated)
+- Unit tests: doctor/bootstrap parsing + command dispatch, against fixtures (never modifying the real repo state)
 
 **Success criteria:**
-- A sandbox can be created, written to, snapshotted, and destroyed via API
-- Isolation verified: a sandbox cannot read another learner's data
-- Resource limits enforced and observable — enforcement mechanism: rlimits (memory/CPU) + wall-clock reaper + workdir-size sweep; per-sandbox pids and hard-disk-quota are accepted v0.3 gaps (no cgroup delegation/sudo on this box, G-1)
+- `nextcraft doctor` reports each prerequisite with actionable guidance
+- `nextcraft bootstrap` on a fresh clone reaches a state where `verify` passes
+- All commands have `--help`, exit non-zero on failure, no shell-out without timeout
+- `pnpm build`, `pnpm typecheck`, `pnpm ai:test` green
 
 ---
 
-### Phase 2: Live Build Telemetry
+### Phase 2: Binary Build + Release Pipeline
 
-**Goal:** Capture in-environment process events and stream them to ai-service reliably.
+**Goal:** Reproducible linux x64 binary + one-liner install + release-asset upload wired into the ship flow.
 
-**Requirements:** REQ-3-003
+**Requirements:** REQ-4-003, REQ-4-004
 
 **Key deliverables:**
-- Telemetry capture agent (in-sandbox): commands, file diffs, run/test results, keystroke-level/activity events
-- Telemetry transport: durable, ordered, resumable stream to ai-service ingestion endpoint
-- Trace persistence: per-learner, per-task process traces stored for grading and proctoring
-- Transport hardening: retries, backpressure, exactly-once-or-at-least-once semantics documented
+- Build script producing `nextcraft-linux-x64` + `nextcraft-linux-x64.sha256` (toolchain probe-verified at RESEARCH; embedded script assets)
+- One-liner install script `install.sh` served from the repo: detect linux x64, resolve latest release via Gitea API, download + verify checksum, install to `~/.local/bin`, PATH hint, source-bootstrap fallback when no binary/asset
+- Ship integration: every release from v0.4 onward attaches the binary + checksum as release assets (the "ongoing binaries" requirement)
+- Asset-upload helper using the Gitea token from `.env*` files only (never shell env)
 
 **Success criteria:**
-- Sandbox activity produces a complete ordered process trace in ai-service
-- Stream survives transient network failure without trace loss
-- Trace retrievable by learner+task ID for grading
+- Binary runs on this box: `./nextcraft-linux-x64 doctor` green against the repo
+- Install script verified end-to-end against the real Gitea release (or local dry-run if release pending)
+- Checksum verification rejects a corrupted download (tested)
+- Release assets present on the phase ship
 
 ---
 
-### Phase 3: Process-Trace Grading Engine
+### Phase 3: Install Docs + Fresh-Clone E2E
 
-**Goal:** Grade learner artifacts from their full process traces.
+**Goal:** Documentation and end-to-end proof that a fresh consumer reaches a running stack via the one-liner.
 
-**Requirements:** REQ-3-004
+**Requirements:** REQ-4-005
 
 **Key deliverables:**
-- Trace analyzer: reconstructs build/decision timeline from a process trace
-- Grading engine: rubric-aligned scoring over the trace (process quality, not just final artifact)
-- Structured score output consumable by the Assessor agent
-- Calibration against v0.2 mock corpora to validate grading dimensions
+- README quickstart: one-liner → `nextcraft doctor` → `nextcraft bootstrap` → `nextcraft dev`
+- CLI command reference (all flags, exit codes)
+- Fresh-clone E2E test: clean temp clone → doctor → bootstrap → verify → build green (sandboxed; no network beyond package registries already used)
+- Install-script docs: prerequisites, offline/manual install, troubleshooting
 
 **Success criteria:**
-- Engine emits structured rubric-aligned scores from a real process trace
-- Scores distinguish process quality (e.g., iterative debugging vs. paste-and-run)
-- Output feeds Assessor; replaces pre-baked artifact corpus inputs
+- A fresh clone bootstraps to a passing `verify` with one command sequence
+- README quickstart matches the actual tested flow exactly
+- E2E test green in CI-equivalent local run
 
 ---
 
-### Phase 4: Variant Task Generation
+### Phase 4: Final Review + Ship
 
-**Goal:** Generate per-learner task variants so no two learners receive identical prompts.
-
-**Requirements:** REQ-3-005
-
-**Key deliverables:**
-- Variant generator: parameterized task templates → unique per-learner instances
-- Variant seed registry: record variant parameters for grading fairness and proctoring
-- Difficulty normalization: variants calibrated to equivalent difficulty
-
-**Success criteria:**
-- Two learners requesting the same competency receive distinct task variants
-- Variant parameters persisted and auditable
-- Grading engine scores variants equitably
-
----
-
-### Phase 5: Oral / Voice Defense
-
-**Goal:** AI examiner conducts a spoken defense of the learner's submitted work.
-
-**Requirements:** REQ-3-006
-
-**Key deliverables:**
-- Voice pipeline: STT → defense dialogue (LLM examiner) → TTS
-- Examiner agent: probes understanding, challenges process choices from the trace
-- Transcript + integrity signals captured for Proctor/Mentor
-- Latency budget: defense feels conversational (bounded turn latency)
-
-**Success criteria:**
-- A spoken defense runs end-to-end (speak → examiner question → learner response → verdict)
-- Transcript + integrity signals persisted and consumable by Proctor
-- Turn latency within the documented budget
-
----
-
-### Phase 6: Agent Re-grounding + Learner Surface Integration
-
-**Goal:** Move Lab/Assessor/Proctor to real engine inputs; wire learner surface to the real engines.
-
-**Requirements:** REQ-3-007, REQ-3-008
-
-**Key deliverables:**
-- Lab agent consumes live sandbox telemetry (replaces v0.2 mock telemetry)
-- Assessor agent consumes grading-engine output (replaces pre-baked artifacts)
-- Proctor consumes telemetry + defense integrity signals (replaces mock telemetry)
-- Learner sandbox mockup → real in-browser build/run; assessment mockup → live defense + live grading
-
-**Success criteria:**
-- Lab/Assessor/Proctor operate on real inputs with no mock fallback in the learner path
-- Learner can build in-browser and see live telemetry + live feedback
-- Assessment surface runs a live defense and shows live grading
-- `pnpm build` and `pnpm typecheck` pass
-
----
-
-### Phase 7: Final Review + Ship
-
-**Goal:** Code review, audit, milestone release.
+**Goal:** Code review, audit, milestone release with binary assets.
 
 **Key deliverables:**
 - Multi-persona code review (correctness, testing, security, performance, maintainability)
 - Project health audit (reconstruction test, .ciagent/ file discipline, branch hygiene, commit discipline)
-- Milestone ship: merge milestone → main, tag final v0.2.x patch, create Gitea release
+- Milestone ship: merge milestone → main, tag final v0.3.x patch, create Gitea release WITH binary + checksum assets, verify assets downloadable
 
 **Success criteria:**
 - Code review: P0 fixes applied, P1+ documented
 - Audit: all checks pass, project state reconstructable from git log
-- Ship: milestone tagged, branch merged to main, Gitea release created — release note states identity/age-gating (KYC) is deferred and age-gating remains a visual mockup
-- All v0.3 requirements marked complete
+- Ship: milestone tagged, branch merged to main, Gitea release created with `nextcraft-linux-x64` + `.sha256` assets attached — the first of the ongoing binary releases
 
 ---
 
+## v0.4 (Complete — Shipped as v0.3.4)
+
+Distribution & Bootstrap CLI: `nextcraft` CLI (doctor/bootstrap/verify/dev) as a self-contained linux x64 SEA binary, one-liner install with sha256 + version integrity gates, release-asset pipeline attaching binaries to every ongoing release (v0.3.2 onward), install/quickstart docs backed by a fresh-clone E2E test. 5 phases (P0–P4). All 5 requirements (REQ-4-001..005) complete. Tags v0.3.0–v0.3.3 per phase, milestone release v0.3.4.
+
+## v0.3 (Complete — Shipped as v0.2.8)
+
+Credential Engines: real sandbox fabric (Linux namespaces), live build telemetry
+(at-least-once/exactly-once), process-trace grading (G-4 gated), seeded per-learner
+variants (fairness anchors wired to grading), oral defense with integrity signals
+(mock-first voice, browser fallback), and real learner build/defense/grading surfaces.
+8 phases. All 8 requirements (REQ-3-001..008) complete. Tags v0.2.1–v0.2.7 per phase,
+milestone release v0.2.8.
+
 ## v0.2 (Complete — Shipped as v0.2.0)
 
 AI Tutor Architecture: Six AI tutor agents (Coach, Tutor, Lab, Assessor, Proctor, Mentor) as real LLM-backed services over mock engine inputs, wired into the learner surface with streaming. 7 phases. All 12 requirements complete. Milestone release v0.2.0.

.ciagent/config.json

@@ -46,9 +46,9 @@
   "projects": [],
   "active_project": null,
   "milestone": {
-    "version": "v0.3",
-    "name": "credential-engines",
+    "version": "v0.4",
+    "name": "distribution",
     "type": "feature",
-    "branch": "milestone/v0.3-credential-engines"
+    "branch": "milestone/v0.4-distribution"
   }
 }

.gitignore

@@ -55,3 +55,6 @@ apps/ai-service/ai_service/data/
 apps/ai-service/**/sandboxes/
 *.db
 *.db-journal
+
+# in-sandbox capture agent runtime spool
+.nc-agent/

README.md

@@ -2,8 +2,71 @@
 
 AI-native outcome school + marketplace — graduates prove what they can build, not what they can write.
 
+## Quickstart
+
+One-liner install (linux x64):
+
+```sh
+curl -fsSL https://git.coreci.dev/coreci/nextcraft/raw/main/scripts/install.sh | sh
+```
+
+That downloads the latest release's `nextcraft` CLI binary, verifies its sha256 checksum, and installs it to `~/.local/bin` (PATH hint printed if needed). Every release ships fresh binaries — re-run the one-liner to upgrade.
+
+Then, from a clone of this repo:
+
+```sh
+nextcraft doctor      # check prerequisites: node >= 18, pnpm >= 8, python3 >= 3.11, git, unshare
+nextcraft bootstrap   # pnpm install + ai-service venv + .env from template (idempotent)
+nextcraft verify      # health check: venv imports, uvicorn, ports, env
+nextcraft dev         # run the ai-service dev server on :8420 (web dev server: pnpm dev)
+```
+
+The E2E test (`apps/cli/tests/fresh-clone-e2e.test.ts`) proves this exact sequence on a fresh clone.
+
+### No binary / non-linux?
+
+The installer degrades to printed source instructions. Manual equivalent:
+
+```sh
+git clone https://git.coreci.dev/coreci/nextcraft.git && cd nextcraft
+pnpm install
+bash apps/ai-service/scripts/bootstrap.sh
+cp apps/ai-service/.env.example apps/ai-service/.env
+pnpm ai:dev
+```
+
+## CLI reference (`nextcraft`)
+
+| Command | What it does | Exit codes |
+|---------|--------------|------------|
+| `doctor` | Checks prerequisites on PATH: node >= 18, pnpm >= 8, python3 >= 3.11, git, unshare (sandbox fabric). Every ✗ prints a fix hint. | 0 all pass, 1 any fail |
+| `bootstrap` | Sets up the monorepo from a fresh clone: (1) locates the repo root, (2) `pnpm install`, (3) ai-service venv via `apps/ai-service/scripts/bootstrap.sh`, (4) copies `.env.example` → `.env` if absent, (5) warns on missing optional keys. Idempotent — safe to re-run. | 0 ok, 1 step failed |
+| `verify` | Health check: ai-service venv + `import ai_service`, uvicorn importable, `.env` present (warn-only), `AI_PORT` (default 8420) free, workspace `node_modules` present. | 0 ok, 1 failures |
+| `dev` | Thin passthrough to `apps/ai-service/scripts/dev.sh` (exports secrets from `.ciagent/.env.secrets` if present, runs uvicorn on :8420). Ctrl+C stops it. The web dev server is separate: `pnpm dev`. | child's exit code |
+| `--help` / `-h` | Usage for the CLI or any command. | 0 |
+| `--version` | Prints the version this binary was built as (matches the release tag). | 0 |
+
+Exit-code contract: `0` success, `1` check/step failure (hint printed), `2` usage error.
+
+### Remote server
+
+`nextcraft dev` binds the API on **0.0.0.0:8420** (and `pnpm dev` serves the web app on all interfaces), so the stack works from other machines out of the box:
+
+- Browse `http://<your-host>:3000` — the web app targets `http://<your-host>:8420` automatically (derived from the browser's hostname).
+- CORS admits any origin (`AI_CORS_ORIGINS=*` in `apps/ai-service/.env`). This is safe **only** because credentials are never enabled; to restrict, set an explicit list: `AI_CORS_ORIGINS=http://<your-host>:3000`.
+- To revert to loopback-only: `AI_HOST=127.0.0.1` in `apps/ai-service/.env`.
+- Security note: this is an unauthenticated dev API reachable from any network the box exposes. Mitigations that still apply: per-learner sandbox caps + global rate caps + learner allowlist (G-5), telemetry flood control (traces marked `INCOMPLETE_FLOODED` are refused by the grader). Expose only on trusted networks until identity/KYC lands (v0.5).
+
+## Docs
+
+- [apps/cli/README.md](apps/cli/README.md) — CLI internals: build, binary pipeline, troubleshooting
+- [.ciagent/PROJECT.md](.ciagent/PROJECT.md) — product spec and milestone history
+- [.ciagent/ARCHITECTURE.md](.ciagent/ARCHITECTURE.md) — system architecture
+
 ## Status
 
-**Milestone v0.1** — UI/UX Prototype (high-fidelity interactive, all mock data)
+**Milestone v0.4** — Distribution & Bootstrap CLI (one-liner install, `nextcraft` binary releases on every ship)
+
+Prior: v0.3 Credential Engines (shipped v0.2.8) · v0.2 AI Tutor Architecture (v0.2.0) · v0.1 UI/UX Prototype (v0.1.0)
 
 Initialized via CIAgent v0.7.0

apps/ai-service/.env.example

@@ -2,6 +2,13 @@
 # Real keys live in .ciagent/.env.secrets (gitignored) and are exported by scripts/dev.sh
 
 AI_PORT=8420
+# Network mode (v0.3.5, D-038): dev server binds 0.0.0.0 so remote machines can
+# reach the stack. Set to 127.0.0.1 to revert to loopback-only.
+AI_HOST=0.0.0.0
+# CORS + WS-origin policy: '*' (default) admits any origin — safe because
+# credentials are never enabled. Restrict with a comma list, e.g.:
+#   AI_CORS_ORIGINS=http://nextcraft-1:3000
+AI_CORS_ORIGINS=*
 AI_PROVIDER=ollama-cloud
 AI_MODEL=gemma4:31b
 AI_OLLAMA_CLOUD_BASE_URL=https://ollama.com/v1
@@ -24,4 +31,9 @@ AI_SANDBOX_MAX_PER_LEARNER=1
 AI_SANDBOX_CREATES_PER_MIN=10
 
 # Persistence (SQLite)
-AI_DB_PATH=ai_service/data/nextcraft.db
+AI_DB_PATH=ai_service/data/nextcraft.db
+# --- v0.3 Voice (REQ-3-006, D-030) ---
+# 'mock' (default; no key needed — tests/dev) or 'browser' (client-native SR/TTS).
+# Real server STT/TTS ('openai-audio' + AI_VOICE_BASE_URL/AI_VOICE_API_KEY)
+# is deferred to v0.4 per GRILL CUT-1/G-7 — keys never in code or commits.
+AI_VOICE_PROVIDER=mock

apps/ai-service/README.md

@@ -194,6 +194,65 @@ Unprivileged user namespaces are on the box's kernel and need neither a
 daemon, nor suid helpers, nor network access — they are the only isolation
 primitive that works here, so that's what v0.3 uses.
 
+## Telemetry delivery semantics (v0.3, REQ-3-003)
+
+Delivery is **at-least-once**; storage is **exactly-once** — the two compose:
+
+- The in-sandbox capture agent (stdlib-only, `scripts/sandbox-agent.py`)
+  spools every event to a durable JSONL file (fsync per append) BEFORE any
+  send attempt, so no event can be lost to a dead socket or a SIGKILL.
+- The WS ingest endpoint (`WS /v1/telemetry/ingest?learner_id&task_id`,
+  D-026) dedups server-side on the `(learner_id, task_id, seq)` primary key:
+  re-sends (reconnect flushes, replay margin) are collapsed, never upserted.
+- On disconnect the agent reconnects with exponential backoff and flushes
+  the spool in `seq` order; a transient outage therefore loses nothing and
+  stores each event exactly once (`tests/telemetry/test_durability.py`
+  proves this end-to-end against a real namespace sandbox + live server).
+- Replay/read path: `GET /v1/telemetry/traces/{learner}/{task}` returns the
+  complete ordered trace; `GET /v1/telemetry/gaps/{learner}/{task}` returns
+  missing seqs for gap detection.
+- Flood boundary (G-3): a connection exceeding `AI_TELEMETRY_MAX_EVENTS_PER_TASK`
+  (default 50,000) is closed with WS code 1008 and its trace is marked
+  `INCOMPLETE_FLOODED` — a terminal integrity flag the grader refuses to
+  grade. Silent event dropping is forbidden: it would corrupt grading input.
+
+## Voice defense (v0.3, REQ-3-006)
+
+Voice is **mock-first** (D-030): the defense pipeline is fully proven over
+the deterministic `MockVoiceProvider` + browser-native fallback — no task
+requires a real voice key. Real server STT/TTS (`OpenAIAudioProvider` over
+OpenAI-compatible `/audio/transcriptions` + `/audio/speech`) is **deferred
+to v0.4** together with KYC (GRILL CUT-1 / G-7): it could never be exercised
+in CI, so v0.3 ships the protocol seam instead of an unverifiable claim.
+
+- `AI_VOICE_PROVIDER=mock` (default) — deterministic canned STT/TTS
+- `AI_VOICE_PROVIDER=browser` — the web client uses SpeechRecognition +
+  speechSynthesis; the server keeps text-turn persistence
+- Conversational budget: a defense turn should complete in **< 4s**
+  (`DEFENSE_TURN_BUDGET_MS` in `tests/voice/test_latency.py`). v0.3
+  asserts instrumentation (stt_ms/llm_ms/tts_ms populated per turn); the
+  wall-clock acceptance probe against a real voice endpoint is a v0.4
+  criterion, run manually with `AI_VOICE_PROVIDER` set to the real
+  provider and keys in `.ciagent/.env.secrets` (never in code/commits).
+
+## End-to-end credential flow (v0.3, REQ-3-007/008)
+
+`tests/api/test_e2e_credential_flow.py` runs the full pipeline against a REAL
+uvicorn server with REAL namespace sandboxes (mock LLM/voice per G-2
+precedent): variant -> telemetry-wired sandbox -> in-sandbox exec -> trace
+persistence -> process-trace grade (variant seed stamped) -> assessor
+coaching -> oral defense -> verdict + integrity signals -> proctor. It
+asserts no corpus fixture appears anywhere in the learner path.
+
+Manual browser pass (documented, not automated): `pnpm ai:dev` + `pnpm dev`,
+then open `/build/stack-orchestration-c007` — variant statement + starter
+files load, edit a file, Run/Test execute in the sandbox with output in the
+read-only panel, the telemetry status pulses, Lab streams feedback from the
+live digest; then `/defend/stack-orchestration-c007` — Start Defense, typed
+answers (mic path needs permission), Finish, Grade My Work renders the real
+rubric bars. Navigating away destroys the sandbox
+(`curl localhost:8420/v1/sandboxes` shows the count drop).
+
 ## Layout
 
 ```

apps/ai-service/ai_service/agents/assessor.py

@@ -1,53 +1,34 @@
-"""AssessorAgent — rubric application to pre-baked artifacts (REQ-2-008).
+"""AssessorAgent — rubric coaching over REAL grading output (REQ-3-007).
 
-Structured-output showcase: applies the 4-layer defense (D-020) to return
-a pydantic-validated rubric score. Mock engine inputs (corpus artifacts +
-transcripts); real process-trace grading is v0.3+.
+v0.3 re-grounding: the Assessor no longer invents scores from corpus
+artifacts — the process-trace grading engine (Phase 3) computes and
+persists the validated RubricScore. This agent now renders the STORED
+grade as rubric-anchored coaching: explains the criteria, cites strengths
+and gaps, and frames next steps. Corpus artifacts are retired from this
+path (corpus dormancy, Task 6-1-04).
 """
 
 from pydantic import BaseModel, Field
 
-from ..corpus.artifacts import (
-    ArtifactSubmission,
-    AssessmentRubric,
-    DefenseTranscript,
-    render_rubric,
-    render_transcript,
-)
 from ..corpus.learner_context import LearnerContext, get_learner_context
+from ..grading.store import GradeRecord
 from ..prompts.assessor import SYSTEM_PROMPT, render_context
 from .base import BaseAgent
 
 
-class CriterionScore(BaseModel):
-    criterion_id: str
-    name: str
-    score: int = Field(ge=0, le=100)
-    evidence: str
+class GradeCoaching(BaseModel):
+    """Rubric-anchored coaching rendered FROM the stored grade (not invented)."""
+
+    summary: str = Field(min_length=1)
+    strengths: list[str] = Field(min_length=1, max_length=3)
+    gaps: list[str] = Field(min_length=1, max_length=3)
+    next_steps: list[str] = Field(min_length=1, max_length=3)
 
 
-class RubricScore(BaseModel):
-    rubric_id: str
-    artifact_id: str
-    competency_id: str
-    scores: list[CriterionScore]
-    strengths: list[str] = Field(min_length=1, max_length=2)
-    gaps: list[str] = Field(min_length=1, max_length=2)
-    verdict: str  # "mastered" | "developing" | "not_yet"
-
-    def weighted_total(self, rubric: AssessmentRubric) -> float:
-        by_id = {c.criterion_id: c for c in rubric.criteria}
-        total = 0.0
-        for s in self.scores:
-            total += s.score * by_id[s.criterion_id].weight
-        return total
-
-
-RUBRIC_SCORE_SCHEMA_HINT = (
-    '{"rubric_id": "<id>", "artifact_id": "<id>", "competency_id": "<id>", '
-    '"scores": [{"criterion_id": "<id>", "name": "<name>", "score": <0-100>, '
-    '"evidence": "<one sentence>"}], "strengths": ["<one sentence>"], '
-    '"gaps": ["<one sentence>"], "verdict": "mastered"|"developing"|"not_yet"}'
+GRADE_COACHING_SCHEMA_HINT = (
+    '{"summary": "<two sentences on the grade>", '
+    '"strengths": ["<one sentence>"], "gaps": ["<one sentence>"], '
+    '"next_steps": ["<one sentence>"]}'
 )
 
 
@@ -58,35 +39,25 @@ class AssessorAgent(BaseAgent):
         ctx = learner_context or get_learner_context()
         return SYSTEM_PROMPT.format_map(render_context(ctx))
 
-    def build_evaluation_input(
+    async def coach_grade(
         self,
-        artifact: ArtifactSubmission,
-        rubric: AssessmentRubric,
-        transcript: DefenseTranscript | None,
-    ) -> str:
-        parts = [
-            f"ARTIFACT: {artifact.name} ({artifact.artifact_id})",
-            f"Evidence excerpt: {artifact.evidence_excerpt}",
-            "",
-            render_rubric(rubric),
-        ]
-        if transcript is not None:
-            parts += ["", render_transcript(transcript)]
-        return "\n".join(parts)
-
-    async def evaluate(
-        self,
-        artifact: ArtifactSubmission,
-        rubric: AssessmentRubric,
-        transcript: DefenseTranscript | None,
+        grade: GradeRecord,
         learner_context: LearnerContext | None = None,
-    ) -> RubricScore:
-        evaluation_input = self.build_evaluation_input(artifact, rubric, transcript)
-        score: RubricScore = await self.structured_reply(
+    ) -> GradeCoaching:
+        """Render the STORED grade as coaching via the D-020 defense."""
+        grade_json = {
+            "verdict": grade.verdict,
+            "scores": grade.scores,
+            "digest": grade.digest,
+        }
+        coaching: GradeCoaching = await self.structured_reply(
             history=None,
-            user_input=evaluation_input,
+            user_input=(
+                "The learner's process-trace grade (computed by the grading "
+                f"engine) is:\n{grade_json!r}\nExplain it as coaching."
+            ),
             learner_context=learner_context,
-            schema=RubricScore,
-            schema_hint=RUBRIC_SCORE_SCHEMA_HINT,
+            schema=GradeCoaching,
+            schema_hint=GRADE_COACHING_SCHEMA_HINT,
         )
-        return score
+        return coaching

apps/ai-service/ai_service/agents/examiner.py

@@ -0,0 +1,104 @@
+"""ExaminerAgent — the seventh agent: oral-defense examiner (REQ-3-006, A-109).
+
+BOUNDARY DECISION (PERSONAS conflict rule, honored by construction): the
+examiner is a TEXT agent. It composes the LLM provider through BaseAgent and
+consumes defense transcript turns; it NEVER imports voice/ — STT/TTS belong
+to the API endpoints (they move audio bytes; the agent moves question text).
+Integrity signals (long pauses, off-scope cadence) are computed by the
+endpoint layer from turn metadata (latency_ms etc.), not by the agent.
+
+Digest discipline (D-028 mirror): questions are grounded in the compact
+TraceDigest + variant statement — never the raw trace, never learner ids.
+"""
+
+from __future__ import annotations
+
+from pydantic import BaseModel, ConfigDict, Field
+
+from ..grading.features import TraceDigest
+from ..llm.types import Message
+from ..prompts.examiner import SYSTEM_PROMPT, VERDICT_SCHEMA_HINT, render_digest_context
+from .base import BaseAgent
+
+
+class DefenseVerdict(BaseModel):
+    """D-20-validated final defense verdict (structured mode)."""
+
+    model_config = ConfigDict(extra="forbid")
+
+    verdict: str = Field(pattern="^(mastered|developing|not_yet)$")
+    understanding: str = Field(min_length=1)
+    process_justification: str = Field(min_length=1)
+    communication: str = Field(min_length=1)
+    strengths: list[str] = Field(min_length=1, max_length=2)
+    gaps: list[str] = Field(min_length=1, max_length=2)
+
+
+class ExaminerAgent(BaseAgent):
+    """Conducts the oral defense: next_question + final_verdict."""
+
+    name = "examiner"
+
+    def system_prompt(self, learner_context=None) -> str:  # noqa: ANN001
+        """Examiner is context-free (digest-anonymous, D-028 mirror)."""
+        return SYSTEM_PROMPT
+
+    def build_defense_messages(
+        self,
+        trace_digest: TraceDigest | None = None,
+        variant_statement: str | None = None,
+        history: list[Message] | None = None,
+    ) -> list[Message]:
+        """System + grounding + defense transcript (no learner id — D-028)."""
+        digest_json = (
+            trace_digest.model_dump_json() if trace_digest is not None else "{}"
+        )
+        messages: list[Message] = [
+            Message(role="system", content=SYSTEM_PROMPT),
+            Message(role="user", content=render_digest_context(digest_json, variant_statement)),
+            Message(
+                role="assistant",
+                content="Understood. I will question the learner about this build session.",
+            ),
+        ]
+        for m in history or []:
+            messages.append(m)
+        return messages
+
+    async def next_question(
+        self,
+        history: list[Message],
+        trace_digest: TraceDigest | None = None,
+        variant_statement: str | None = None,
+    ) -> str:
+        """One examiner question (streamed over SSE by the endpoints)."""
+        messages = self.build_defense_messages(trace_digest, variant_statement, history)
+        messages.append(
+            Message(role="user", content="Ask the learner your next question now.")
+        )
+        reply = await self.provider.chat(messages, model=self.settings.model)
+        return reply
+
+    async def final_verdict(
+        self,
+        history: list[Message],
+        trace_digest: TraceDigest | None = None,
+        variant_statement: str | None = None,
+    ) -> DefenseVerdict:
+        """Structured verdict via the D-020 4-layer defense."""
+        from .structured import structured_completion  # module-direct (G-4)
+
+        messages = self.build_defense_messages(trace_digest, variant_statement, history)
+        messages.append(
+            Message(
+                role="user",
+                content="The defense is finished. Return the final verdict JSON now.",
+            )
+        )
+        return await structured_completion(
+            self.provider,
+            messages,
+            model=self.settings.model,
+            schema=DefenseVerdict,
+            schema_hint=VERDICT_SCHEMA_HINT,
+        )

apps/ai-service/ai_service/agents/lab.py

@@ -1,17 +1,18 @@
-"""LabAgent — in-flow feedback over simulated sandbox telemetry (REQ-2-007).
+"""LabAgent — in-flow feedback over LIVE sandbox telemetry (REQ-3-007).
 
-Scenario-driven: consumes a LabTelemetryScenario from the corpus, renders
-the event timeline into the conversation, streams concrete feedback.
-No session chat — each request is one scenario read.
+v0.3 re-grounding: consumes a TraceDigest computed from the learner's real
+trace (grading/features.compute_digest over TraceStore events) — the v0.2
+corpus scenarios are retired from this path (corpus dormancy, Task 6-1-04).
+No session chat — each request is one live-trace read.
 """
 
 from collections.abc import AsyncIterator
 
 from ..config import Settings
 from ..corpus.learner_context import LearnerContext, get_learner_context
-from ..corpus.telemetry import LabTelemetryScenario, summarize_scenario
+from ..grading.features import TraceDigest
 from ..llm.base import LLMProvider
-from ..prompts.lab import SYSTEM_PROMPT, render_context
+from ..prompts.lab import SYSTEM_PROMPT, render_context, render_digest_timeline
 from .base import BaseAgent
 
 
@@ -27,10 +28,11 @@ class LabAgent(BaseAgent):
 
     async def stream_feedback(
         self,
-        scenario: LabTelemetryScenario,
+        digest: TraceDigest | None,
         learner_context: LearnerContext | None = None,
     ) -> AsyncIterator[str]:
-        timeline = summarize_scenario(scenario)
+        """Feedback grounded in the learner's live trace digest."""
+        timeline = render_digest_timeline(digest)
         async for token in self.stream_reply(
             history=None, user_input=timeline, learner_context=learner_context
         ):

apps/ai-service/ai_service/agents/proctor.py

@@ -1,33 +1,41 @@
-"""ProctorAgent — integrity signals + coaching interventions (REQ-2-009).
+"""ProctorAgent — integrity signals + coaching over REAL inputs (REQ-3-007).
 
-Consumes a ProctorScenario from the corpus, returns pydantic-validated
-signal classifications via structured_reply (4-layer defense).
-Mock engine inputs; real identity/attention signals are v0.3+.
+v0.3 re-grounding: consumes the learner's live trace digest (idle gaps,
+command cadence), the DefenseStore integrity signals (long pauses from the
+oral defense), and the variant seed cross-check — NOT v0.2 corpus
+scenarios. The proctor COACHES: it classifies signals supportively and
+recommends one intervention; it never punishes and never accuses.
+
+Integrity inputs (computed server-side, passed in by the API layer):
+- trace digest: idle_gap_count/total, command_categories histogram,
+  error/fix cycles, huge-burst indicators (edit_count vs test runs)
+- defense signals: long_pauses list from the finished defense (A-109)
+- variant: seed + params when the task is variant-derived (off-template
+  work is a cross-check input, not an accusation)
 """
 
 from pydantic import BaseModel, Field
 
 from ..corpus.learner_context import LearnerContext, get_learner_context
-from ..corpus.telemetry import ProctorScenario, summarize_proctor_scenario
+from ..grading.features import TraceDigest
 from ..prompts.proctor import SYSTEM_PROMPT, render_context
 from .base import BaseAgent
 
 
 class IntegritySignal(BaseModel):
-    signal_type: str  # e.g. "context_switch" | "idle_gap" | "large_paste"
+    signal_type: str  # "idle_gap" | "long_pause" | "burst_edit" | "off_template"
     severity: str  # "low" | "medium" | "high"
     note: str
 
 
 class ProctorAssessment(BaseModel):
-    scenario_id: str
     signals: list[IntegritySignal] = Field(min_length=0)
     intervention: str  # ONE supportive coaching recommendation
     summary: str
 
 
 PROCTOR_ASSESSMENT_SCHEMA_HINT = (
-    '{"scenario_id": "<id>", "signals": [{"signal_type": "<type>", '
+    '{"signals": [{"signal_type": "<type>", '
     '"severity": "low"|"medium"|"high", "note": "<one sentence>"}], '
     '"intervention": "<one supportive recommendation>", '
     '"summary": "<one sentence>"}'
@@ -43,13 +51,27 @@ class ProctorAgent(BaseAgent):
 
     async def assess(
         self,
-        scenario: ProctorScenario,
+        digest: TraceDigest | None,
+        defense_signals: dict | None = None,
+        variant_context: dict | None = None,
         learner_context: LearnerContext | None = None,
     ) -> ProctorAssessment:
-        timeline = summarize_proctor_scenario(scenario)
+        """Classify REAL integrity inputs into supportive signals + coaching."""
+        parts: list[str] = []
+        if digest is not None:
+            parts.append(f"Build-session digest:\n{digest.model_dump_json()}")
+        else:
+            parts.append("No build telemetry recorded for this task yet.")
+        if defense_signals:
+            parts.append(f"Oral-defense integrity signals:\n{defense_signals}")
+        if variant_context:
+            parts.append(f"Variant audit context (seed + params):\n{variant_context}")
         assessment: ProctorAssessment = await self.structured_reply(
             history=None,
-            user_input=timeline,
+            user_input=(
+                "Assess this learner's integrity signals supportively.\n\n"
+                + "\n\n".join(parts)
+            ),
             learner_context=learner_context,
             schema=ProctorAssessment,
             schema_hint=PROCTOR_ASSESSMENT_SCHEMA_HINT,

apps/ai-service/ai_service/agents/registry.py

@@ -14,13 +14,14 @@ AgentFactory = Callable[[LLMProvider, Settings], BaseAgent]
 
 
 def register_builtin_agents(registry: "AgentRegistry") -> None:
-    """Central registration of all six shipped tutor agents (G-4: one pattern).
+    """Central registration of all seven shipped agents (G-4: one pattern).
 
-    coach, tutor, lab, assessor, proctor, mentor. New agents register here
-    in their landing phase.
+    coach, tutor, lab, assessor, proctor, mentor, examiner (Phase 5).
+    New agents register here in their landing phase.
     """
     from .assessor import AssessorAgent
     from .coach import CoachAgent
+    from .examiner import ExaminerAgent
     from .lab import LabAgent
     from .mentor import MentorAgent
     from .proctor import ProctorAgent
@@ -38,6 +39,9 @@ def register_builtin_agents(registry: "AgentRegistry") -> None:
     registry.register(
         "mentor", lambda provider, settings: MentorAgent(provider, settings)
     )
+    registry.register(
+        "examiner", lambda provider, settings: ExaminerAgent(provider, settings)
+    )
 
 
 class UnknownAgentError(KeyError):

apps/ai-service/ai_service/api/__init__.py

@@ -5,10 +5,13 @@ Boundary rule: api/ composes agents/ and llm/; they never import api/.
 
 from .assessment import router as assessment_router
 from .chat import router as chat_router
+from .defense import router as defense_router
 from .lab import router as lab_router
 from .mentor import router as mentor_router
 from .proctor import router as proctor_router
 from .sandboxes import router as sandboxes_router
+from .telemetry import router as telemetry_router
+from .variants import router as variants_router
 
 __all__ = [
     "assessment_router",
@@ -17,4 +20,7 @@ __all__ = [
     "mentor_router",
     "proctor_router",
     "sandboxes_router",
+    "telemetry_router",
+    "variants_router",
+    "defense_router",
 ]

apps/ai-service/ai_service/api/assessment.py

@@ -1,47 +1,190 @@
-"""POST /v1/assessment/evaluate — structured rubric scores (REQ-2-008).
+"""/v1/assessment — rubric evaluation + trace grading endpoints (REQ-2-008, REQ-3-004).
 
-JSON response (not SSE): a pydantic-validated RubricScore. Unknown
-artifact → 404. The Assessor's structured output IS the payload.
+Two endpoint families share this router:
+
+    POST /v1/assessment/evaluate                  (v0.2, REQ-2-008) — corpus
+                                                 artifact evaluation through
+                                                 the Assessor agent.
+    POST /v1/assessment/grade                     (v0.3, REQ-3-004) — grade a
+                                                 REAL process trace through
+                                                 the GradingEngine.
+    GET  /v1/assessment/grade/{learner_id}/{task_id} — stored latest grade.
+
+Grading status-code mapping (the engine's outcomes are CONTRACT, not errors):
+
+    GradeRecord(verdict=GRADED)                    → 200 — rubric scores +
+                                                    verdict (in scores.verdict)
+                                                    + digest summary.
+    GradeRecord(UNGRADABLE_TRACE_INCOMPLETE)       → 200 — the ungradable
+                                                    record IS a valid result:
+                                                    the trace cannot be graded,
+                                                    and the gate surfaces WHY
+                                                    (scores.missing_seqs +
+                                                    scores.integrity_flag).
+                                                    Persisted like any grade.
+    GradeRecord(UNGRADABLE_EMPTY_TRACE)           → 200 — no events stored for
+                                                    the pair. This covers BOTH
+                                                    a known pair whose trace
+                                                    ended up empty AND a task
+                                                    that never had a trace at
+                                                    all: the engine cannot
+                                                    distinguish them (zero
+                                                    stored events is zero
+                                                    events), and grading an
+                                                    absent trace genuinely has
+                                                    the empty-trace outcome —
+                                                    a 404 here would erase the
+                                                    durable gate record the
+                                                    engine persists for the
+                                                    pair. PLAN's 404 applies to
+                                                    GET of a never-graded pair.
+    StructuredOutputError                          → 502 — the trace was
+                                                    gradable but the provider
+                                                    failed the D-020 budget;
+                                                    provider failure (bad
+                                                    gateway to the model), same
+                                                    mapping as evaluate.
+    GET of an unknown (never-graded) pair          → 404.
+
+DI (D-027/D-032 house pattern): engine + store arrive via deps.get_grading_engine
+/ get_grade_store from app.state; this module owns all FastAPI wiring — the
+engine knows nothing of HTTP. UNGRADABLE_* bodies are rendered by the same
+GradeResponse model as GRADED ones (a gate record's `scores` holds the gate
+detail instead of rubric scores), so consumers read ONE shape.
 """
 
+from datetime import datetime
+from typing import Any
+
 from fastapi import APIRouter, Depends, HTTPException
 from pydantic import BaseModel, Field
 
-from ..agents.assessor import RubricScore
 from ..agents.registry import AgentRegistry
+from ..agents.structured import StructuredOutputError
 from ..config import Settings
-from ..corpus.artifacts import get_artifact_bundle, get_transcript_for_artifact
 from ..corpus.learner_context import get_learner_context
-from .deps import get_agent_registry, get_provider, get_settings
+from ..grading.engine import GradingEngine
+from ..grading.store import GradeRecord, GradeStore
+from .deps import (
+    get_agent_registry,
+    get_grade_store,
+    get_grading_engine,
+    get_provider,
+    get_settings,
+)
 
 router = APIRouter(prefix="/v1")
 
 
-class AssessmentRequest(BaseModel):
-    artifact_id: str = Field(min_length=1)
-    learner_id: str | None = None
+# --- v0.2 artifact evaluation (REQ-2-008) --------------------------------------
 
 
-@router.post("/assessment/evaluate", response_model=RubricScore)
+class EvaluateRequest(BaseModel):
+    learner_id: str = Field(min_length=1)
+    task_id: str = Field(min_length=1)
+
+
+@router.post("/assessment/evaluate")
 async def assessment_evaluate(
-    body: AssessmentRequest,
+    body: EvaluateRequest,
     registry: AgentRegistry = Depends(get_agent_registry),
     settings: Settings = Depends(get_settings),
     provider=Depends(get_provider),
-) -> RubricScore:
-    bundle = get_artifact_bundle(body.artifact_id)
-    if bundle is None:
+    grade_store=Depends(get_grade_store),
+) -> dict:
+    """Assessor coaching rendered FROM the learner's stored grade (REQ-3-007).
+
+    The grading engine computes the scores (POST /assessment/grade); this
+    endpoint explains them. No stored grade yet -> 404 (grade first).
+    """
+    grade = grade_store.get(body.learner_id, body.task_id)
+    if grade is None:
         raise HTTPException(
-            status_code=404, detail=f"unknown artifact {body.artifact_id!r}"
+            status_code=404,
+            detail=f"no stored grade for {body.learner_id}/{body.task_id} - grade first",
         )
-    artifact, rubric = bundle
-    transcript = get_transcript_for_artifact(artifact.artifact_id)
     agent = registry.get(provider, settings, "assessor")
     learner_context = get_learner_context(body.learner_id)
     try:
-        return await agent.evaluate(artifact, rubric, transcript, learner_context)
+        coaching = await agent.coach_grade(grade, learner_context)
     except Exception as exc:
         raise HTTPException(
             status_code=502,
             detail=f"assessment evaluation failed: {exc}",
         ) from exc
+    return {
+        "learner_id": grade.learner_id,
+        "task_id": grade.task_id,
+        "grade_verdict": grade.verdict,
+        "grade_scores": grade.scores,
+        "coaching": coaching.model_dump(),
+    }
+
+
+# --- # --- v0.3 trace grading (REQ-3-004) ---------------------------------------------
+
+
+class GradeRequest(BaseModel):
+    learner_id: str = Field(min_length=1)
+    task_id: str = Field(min_length=1)
+
+
+class GradeResponse(BaseModel):
+    """GradeRecord over HTTP — one shape for GRADED and UNGRADABLE_* alike.
+
+    `scores` holds the validated rubric (criteria 0-4, strengths, gaps,
+    rubric verdict) for a GRADED record, or the gate detail
+    ({integrity_flag, missing_seqs}) for an UNGRADABLE_* record — never both.
+    `digest` is the compact trace summary that fed the rubric prompt (empty
+    for gate records: nothing was graded).
+    """
+
+    learner_id: str
+    task_id: str
+    variant_seed: str | None
+    digest: dict[str, Any]
+    scores: dict[str, Any]
+    verdict: str
+    model: str
+    created_at: datetime
+
+
+def _grade_response(record: GradeRecord) -> GradeResponse:
+    return GradeResponse.model_validate(record, from_attributes=True)
+
+
+@router.post("/assessment/grade", response_model=GradeResponse)
+async def assessment_grade(
+    body: GradeRequest,
+    engine: GradingEngine = Depends(get_grading_engine),
+) -> GradeResponse:
+    """Run the grading engine for one (learner_id, task_id) trace.
+
+    Gate outcomes (UNGRADABLE_*) are 200s — they are first-class results the
+    engine persists, not failures. Only a provider that exhausts the D-020
+    budget turns into a 502; nothing is persisted on that path.
+    """
+    try:
+        record = await engine.grade(body.learner_id, body.task_id)
+    except StructuredOutputError as exc:
+        raise HTTPException(
+            status_code=502,
+            detail=f"grading failed: {exc}",
+        ) from exc
+    return _grade_response(record)
+
+
+@router.get("/assessment/grade/{learner_id}/{task_id}", response_model=GradeResponse)
+async def assessment_get_grade(
+    learner_id: str,
+    task_id: str,
+    store: GradeStore = Depends(get_grade_store),
+) -> GradeResponse:
+    """Latest stored grade for the pair; 404 when none was ever stored."""
+    record = store.get(learner_id, task_id)
+    if record is None:
+        raise HTTPException(
+            status_code=404,
+            detail=f"no stored grade for {learner_id!r}/{task_id!r}",
+        )
+    return _grade_response(record)

apps/ai-service/ai_service/api/defense.py

@@ -0,0 +1,332 @@
+"""Oral-defense endpoints (Task 5-3-01, REQ-3-006, A-109).
+
+Full defense loop over HTTP with mock-first voice (D-030) and the seventh
+Examiner agent (SSE question streaming happens through the chat pipeline;
+these endpoints are the session orchestration + transcript persistence):
+
+    POST /v1/defense/start          {learner_id, task_id}
+    POST /v1/defense/{id}/answer    {text} | multipart audio (STT)
+    GET  /v1/defense/{id}/audio/{turn_id}      TTS bytes (streaming)
+    POST /v1/defense/{id}/finish    verdict + integrity signals
+    GET  /v1/defense/{id}           transcript + signals
+
+Integrity signals (A-109) are computed server-side from turn metadata:
+long pauses = learner turns whose latency_ms exceeds PAUSE_THRESHOLD_MS.
+The defense does NOT gate on trace completeness (the grader does, G-4);
+an incomplete trace is surfaced as `trace_complete: false` so the UI can
+disclose it before the learner defends.
+"""
+
+from __future__ import annotations
+
+import time
+from datetime import UTC, datetime
+
+from fastapi import APIRouter, Depends, File, Form, HTTPException, UploadFile
+from fastapi.responses import StreamingResponse
+from pydantic import BaseModel, Field
+
+from ..agents.examiner import ExaminerAgent
+from ..grading.features import TraceDigest, compute_digest
+from ..llm.types import Message
+from ..voice.base import VoiceDescriptor
+from ..voice.browser import BROWSER_FALLBACK_DESCRIPTOR
+from ..voice.defense_store import DefenseRecord, DefenseStore, DefenseTurn
+from .deps import (
+    get_examiner,
+    get_settings,
+    get_trace_store,
+    get_variant_store,
+    get_voice_provider,
+    get_voice_store,
+)
+
+router = APIRouter(prefix="/v1/defense", tags=["defense"])
+
+#: A-109: learner turns slower than this are flagged as long pauses (ms).
+PAUSE_THRESHOLD_MS = 15_000
+
+_ROLE_EXAMINER = "examiner"
+_ROLE_LEARNER = "learner"
+
+
+class StartRequest(BaseModel):
+    learner_id: str = Field(min_length=1)
+    task_id: str = Field(min_length=1)
+
+
+class StartResponse(BaseModel):
+    defense_id: str
+    voice_descriptor: dict
+    trace_complete: bool
+    first_question: str
+
+
+class AnswerResponse(BaseModel):
+    question: str
+    turn_latency: dict[str, int | None]
+
+
+class FinishResponse(BaseModel):
+    verdict: dict
+    integrity_signals: dict
+
+
+async def _digest_for_task(
+    trace_store, learner_id: str, task_id: str
+) -> tuple[TraceDigest | None, bool]:
+    """Digest of the learner's trace for this task + completeness flag."""
+    if not trace_store.list_tasks(learner_id) or task_id not in trace_store.list_tasks(
+        learner_id
+    ):
+        return None, True  # no trace at all is "complete" for defense purposes
+    trace = trace_store.get_trace(learner_id, task_id)
+    gaps = trace_store.gaps(learner_id, task_id)
+    return (compute_digest(trace) if trace else None), (len(gaps) == 0)
+
+
+def _voice_descriptor(settings) -> VoiceDescriptor:
+    """The capability descriptor for the configured voice mode (D-030).
+
+    Must-Have #6: browser mode returns BROWSER_FALLBACK_DESCRIPTOR so the
+    web client selects native SpeechRecognition/speechSynthesis; mock mode
+    returns the mock descriptor. (A v0.4 server provider would return
+    mode="server" — the protocol seam.)
+    """
+    if (settings.voice_provider or "mock").strip().lower() == "browser":
+        return BROWSER_FALLBACK_DESCRIPTOR
+    return VoiceDescriptor(
+        mode="mock", sr_available=True, tts_available=True, hint=""
+    )
+
+
+@router.post("/start", response_model=StartResponse)
+async def start_defense(
+    body: StartRequest,
+    examiner: ExaminerAgent = Depends(get_examiner),
+    voice_store: DefenseStore = Depends(get_voice_store),
+    voice_provider=Depends(get_voice_provider),
+    trace_store=Depends(get_trace_store),
+    variant_store=Depends(get_variant_store),
+    settings=Depends(get_settings),
+) -> StartResponse:
+    record = voice_store.start(
+        DefenseRecord(
+            id=f"dfn-{int(time.time() * 1000):x}-{body.learner_id[:8]}",
+            learner_id=body.learner_id,
+            task_id=body.task_id,
+            status="in_progress",
+            created_at=datetime.now(UTC),
+        )
+    )
+    digest, trace_complete = await _digest_for_task(trace_store, body.learner_id, body.task_id)
+    variant = variant_store.get_by_task(body.task_id)
+    statement = variant.statement if variant is not None else None
+
+    started = time.perf_counter()
+    question = await examiner.next_question(
+        history=[], trace_digest=digest, variant_statement=statement
+    )
+    llm_ms = int((time.perf_counter() - started) * 1000)
+    voice_store.append_turn(
+        record.id,
+        DefenseTurn(
+            defense_id=record.id,
+            seq=0,
+            role=_ROLE_EXAMINER,
+            text=question,
+            ts=datetime.now(UTC),
+            latency_ms=llm_ms,
+            created_at=datetime.now(UTC),
+        ),
+    )
+    descriptor = getattr(voice_provider, "descriptor", None) or _voice_descriptor(
+        settings
+    )
+    return StartResponse(
+        defense_id=record.id,
+        voice_descriptor=descriptor.model_dump(),
+        trace_complete=trace_complete,
+        first_question=question,
+    )
+
+
+@router.post("/{defense_id}/answer", response_model=AnswerResponse)
+async def answer_defense(
+    defense_id: str,
+    text: str | None = Form(default=None),
+    audio: UploadFile | None = File(default=None),
+    voice_store: DefenseStore = Depends(get_voice_store),
+    voice_provider=Depends(get_voice_provider),
+    examiner: ExaminerAgent = Depends(get_examiner),
+    trace_store=Depends(get_trace_store),
+    variant_store=Depends(get_variant_store),
+) -> AnswerResponse:
+    record = voice_store.get(defense_id)
+    if record is None:
+        raise HTTPException(status_code=404, detail=f"no defense {defense_id!r}")
+    if record.status == "finished":
+        # The store owns the finished transition but does NOT police turn
+        # sequencing (defense_store.py: "turns after finalize are a sequencing
+        # bug for the endpoints to prevent") — this is the endpoint half of
+        # that contract: a sealed transcript is append-only-no-more.
+        raise HTTPException(
+            status_code=409, detail="defense is finished; start a new defense"
+        )
+    if text is None and audio is None:
+        raise HTTPException(status_code=422, detail="provide {text} or audio")
+
+    # STT (typed fallback bypasses the voice provider entirely).
+    stt_ms: int | None = None
+    if audio is not None:
+        stt_started = time.perf_counter()
+        raw = await audio.read()
+        if not raw:
+            # Empty upload is a client error (422), not a provider crash
+            # (500): validate before the provider call so every provider —
+            # mock today, the v0.4 real one — sees the same contract.
+            raise HTTPException(status_code=422, detail="audio upload is empty")
+        fmt = (audio.content_type or "audio/wav").split("/")[-1]
+        segment = await voice_provider.transcribe(raw, fmt)
+        stt_ms = int((time.perf_counter() - stt_started) * 1000)
+        text = segment.text
+
+    turns = record.turns if hasattr(record, "turns") else []
+    history = [
+        Message(role="assistant" if t.role == _ROLE_EXAMINER else "user", content=t.text)
+        for t in turns
+    ]
+    next_seq = len(turns)
+
+    voice_store.append_turn(
+        defense_id,
+        DefenseTurn(
+            defense_id=defense_id,
+            seq=next_seq,
+            role=_ROLE_LEARNER,
+            text=text or "",
+            ts=datetime.now(UTC),
+            latency_ms=stt_ms,
+            created_at=datetime.now(UTC),
+        ),
+    )
+
+    digest, _ = await _digest_for_task(trace_store, record.learner_id, record.task_id)
+    variant = variant_store.get_by_task(record.task_id)
+
+    llm_started = time.perf_counter()
+    question = await examiner.next_question(
+        history=history + [Message(role="user", content=text or "")],
+        trace_digest=digest,
+        variant_statement=variant.statement if variant is not None else None,
+    )
+    llm_ms = int((time.perf_counter() - llm_started) * 1000)
+
+    voice_store.append_turn(
+        defense_id,
+        DefenseTurn(
+            defense_id=defense_id,
+            seq=next_seq + 1,
+            role=_ROLE_EXAMINER,
+            text=question,
+            ts=datetime.now(UTC),
+            latency_ms=llm_ms,
+            created_at=datetime.now(UTC),
+        ),
+    )
+    return AnswerResponse(
+        question=question,
+        turn_latency={"stt_ms": stt_ms, "llm_ms": llm_ms, "tts_ms": None},
+    )
+
+
+@router.get("/{defense_id}/audio/{turn_id}")
+async def defense_audio(
+    defense_id: str,
+    turn_id: int,
+    voice_store: DefenseStore = Depends(get_voice_store),
+    voice_provider=Depends(get_voice_provider),
+):
+    record = voice_store.get(defense_id)
+    if record is None:
+        raise HTTPException(status_code=404, detail=f"no defense {defense_id!r}")
+    turn = next((t for t in record.turns if t.seq == turn_id), None)
+    if turn is None or turn.role != _ROLE_EXAMINER:
+        raise HTTPException(status_code=404, detail=f"no examiner turn {turn_id!r}")
+
+    async def stream():
+        async for chunk in voice_provider.synthesize(turn.text):
+            yield chunk
+
+    return StreamingResponse(stream(), media_type="audio/wav")
+
+
+@router.post("/{defense_id}/finish", response_model=FinishResponse)
+async def finish_defense(
+    defense_id: str,
+    voice_store: DefenseStore = Depends(get_voice_store),
+    examiner: ExaminerAgent = Depends(get_examiner),
+    trace_store=Depends(get_trace_store),
+    variant_store=Depends(get_variant_store),
+) -> FinishResponse:
+    record = voice_store.get(defense_id)
+    if record is None:
+        raise HTTPException(status_code=404, detail=f"no defense {defense_id!r}")
+
+    turns = record.turns if hasattr(record, "turns") else []
+    history = [
+        Message(role="assistant" if t.role == _ROLE_EXAMINER else "user", content=t.text)
+        for t in turns
+    ]
+    digest, _ = await _digest_for_task(trace_store, record.learner_id, record.task_id)
+    variant = variant_store.get_by_task(record.task_id)
+    verdict = await examiner.final_verdict(
+        history=history,
+        trace_digest=digest,
+        variant_statement=variant.statement if variant is not None else None,
+    )
+
+    signals: dict = {
+        "long_pauses": [
+            {"turn": t.seq, "latency_ms": t.latency_ms}
+            for t in turns
+            if t.role == _ROLE_LEARNER and (t.latency_ms or 0) > PAUSE_THRESHOLD_MS
+        ],
+        "pause_threshold_ms": PAUSE_THRESHOLD_MS,
+        # Must-Have #1: "verdict + transcript persisted" — the verdict is
+        # stored INSIDE integrity_signals so GET /{id} after finish can
+        # re-serve it (the finish response alone would lose it). Signals
+        # are a JSON object dict (DefenseStore.finalize contract), so the
+        # verdict nests under the "verdict" key alongside the A-109
+        # markers the Proctor/Mentor feeds read.
+        "verdict": verdict.model_dump(),
+    }
+    voice_store.finalize(defense_id, signals)
+    return FinishResponse(verdict=verdict.model_dump(), integrity_signals=signals)
+
+
+@router.get("/{defense_id}")
+async def get_defense(
+    defense_id: str,
+    voice_store: DefenseStore = Depends(get_voice_store),
+):
+    record = voice_store.get(defense_id)
+    if record is None:
+        raise HTTPException(status_code=404, detail=f"no defense {defense_id!r}")
+    return {
+        "defense_id": record.id,
+        "learner_id": record.learner_id,
+        "task_id": record.task_id,
+        "status": record.status,
+        "turns": [
+            {
+                "seq": t.seq,
+                "role": t.role,
+                "text": t.text,
+                "ts": t.ts,
+                "latency_ms": t.latency_ms,
+            }
+            for t in record.turns
+        ],
+        "integrity_signals": record.integrity_signals or {},
+    }

apps/ai-service/ai_service/api/deps.py

@@ -2,12 +2,21 @@
 
 from fastapi import Request
 
+from ..agents.examiner import ExaminerAgent
 from ..agents.registry import AgentRegistry
 from ..agents.session import SessionStore
 from ..config import Settings
+from ..grading.engine import GradingEngine
+from ..grading.store import GradeStore
 from ..llm.base import LLMProvider
 from ..sandbox.manager import SandboxManager
 from ..sandbox.workdir import SandboxDir
+from ..telemetry.ingest import TraceIntegrityMap
+from ..telemetry.store import TraceStore
+from ..variants.generator import VariantGenerator
+from ..variants.store import VariantStore
+from ..voice.base import VoiceProvider
+from ..voice.defense_store import DefenseStore
 
 
 def get_settings(request: Request) -> Settings:
@@ -33,3 +42,38 @@ def get_sandbox_manager(request: Request) -> SandboxManager:
 def get_sandbox_test_layout(request: Request) -> SandboxDir | None:
     """Optional test seam (app.state.sandbox_test_layout); always None in prod."""
     return getattr(request.app.state, "sandbox_test_layout", None)
+
+
+def get_trace_store(request: Request) -> TraceStore:
+    return request.app.state.trace_store
+
+
+def get_trace_integrity(request: Request) -> TraceIntegrityMap:
+    return request.app.state.trace_integrity
+
+
+def get_grade_store(request: Request) -> GradeStore:
+    return request.app.state.grade_store
+
+
+def get_grading_engine(request: Request) -> GradingEngine:
+    return request.app.state.grading_engine
+
+
+def get_variant_generator(request: Request) -> VariantGenerator:
+    return request.app.state.variant_generator
+
+
+def get_variant_store(request: Request) -> VariantStore:
+    return request.app.state.variant_store
+
+def get_voice_store(request: Request) -> DefenseStore:
+    return request.app.state.defense_store
+
+
+def get_voice_provider(request: Request) -> VoiceProvider:
+    return request.app.state.voice_provider
+
+
+def get_examiner(request: Request) -> ExaminerAgent:
+    return request.app.state.examiner_agent

apps/ai-service/ai_service/api/lab.py

@@ -1,27 +1,36 @@
-"""POST /v1/lab/feedback — SSE stream of Lab in-flow feedback (REQ-2-007).
+"""POST /v1/lab/feedback — SSE stream of Lab in-flow feedback (REQ-3-007).
 
-D-016 envelope with agent=lab. Unknown scenario → 404 before streaming.
+v0.3 re-grounding: LIVE trace digest. Request carries {learner_id, task_id};
+the digest is computed from the learner's real TraceStore events (D-028)
+and handed to the Lab agent. No corpus scenarios. Empty/unknown trace is NOT
+an error — Lab gets a "no telemetry yet" timeline and coaches the baseline.
+D-016 envelope with agent=lab.
 """
 
 import json
 from collections.abc import AsyncIterator
 
-from fastapi import APIRouter, Depends, HTTPException
+from fastapi import APIRouter, Depends
 from pydantic import BaseModel, Field
 from sse_starlette.sse import EventSourceResponse
 
 from ..agents.registry import AgentRegistry
 from ..config import Settings
 from ..corpus.learner_context import get_learner_context
-from ..corpus.telemetry import get_lab_scenario
-from .deps import get_agent_registry, get_provider, get_settings
+from ..grading.features import compute_digest
+from .deps import (
+    get_agent_registry,
+    get_provider,
+    get_settings,
+    get_trace_store,
+)
 
 router = APIRouter(prefix="/v1")
 
 
 class LabFeedbackRequest(BaseModel):
-    scenario_id: str = Field(min_length=1)
-    learner_id: str | None = None
+    learner_id: str = Field(min_length=1)
+    task_id: str = Field(min_length=1)
 
 
 @router.post("/lab/feedback")
@@ -30,25 +39,27 @@ async def lab_feedback(
     registry: AgentRegistry = Depends(get_agent_registry),
     settings: Settings = Depends(get_settings),
     provider=Depends(get_provider),
+    trace_store=Depends(get_trace_store),
 ) -> EventSourceResponse:
-    scenario = get_lab_scenario(body.scenario_id)
-    if scenario is None:
-        raise HTTPException(
-            status_code=404, detail=f"unknown scenario {body.scenario_id!r}"
-        )
     agent = registry.get(provider, settings, "lab")
     learner_context = get_learner_context(body.learner_id)
+    trace = (
+        trace_store.get_trace(body.learner_id, body.task_id)
+        if body.task_id in trace_store.list_tasks(body.learner_id)
+        else []
+    )
+    digest = compute_digest(trace) if trace else None
 
     async def event_stream() -> AsyncIterator[dict]:
         yield {"event": "message", "data": json.dumps({
             "type": "meta",
             "agent": "lab",
-            "scenario_id": body.scenario_id,
+            "task_id": body.task_id,
             "model": settings.model,
         })}
         first_byte = True
         try:
-            async for token in agent.stream_feedback(scenario, learner_context):
+            async for token in agent.stream_feedback(digest, learner_context):
                 first_byte = False
                 yield {"event": "message", "data": json.dumps({
                     "type": "delta", "content": token

apps/ai-service/ai_service/api/proctor.py

@@ -1,7 +1,8 @@
-"""POST /v1/proctor/signals — structured integrity signals (REQ-2-009).
+"""POST /v1/proctor/signals — integrity signals over REAL inputs (REQ-3-007).
 
-JSON response (not SSE): a pydantic-validated ProctorAssessment.
-Unknown scenario → 404. Coaching-shaped interventions only.
+v0.3 re-grounding: live trace digest + DefenseStore long-pause signals +
+variant seed cross-check, no corpus scenarios. The proctor coaches:
+a pydantic-validated ProctorAssessment (JSON response, not SSE).
 """
 
 from fastapi import APIRouter, Depends, HTTPException
@@ -11,15 +12,22 @@ from ..agents.proctor import ProctorAssessment
 from ..agents.registry import AgentRegistry
 from ..config import Settings
 from ..corpus.learner_context import get_learner_context
-from ..corpus.telemetry import get_proctor_scenario
-from .deps import get_agent_registry, get_provider, get_settings
+from ..grading.features import compute_digest
+from .deps import (
+    get_agent_registry,
+    get_provider,
+    get_settings,
+    get_trace_store,
+    get_variant_store,
+    get_voice_store,
+)
 
 router = APIRouter(prefix="/v1")
 
 
 class ProctorRequest(BaseModel):
-    scenario_id: str = Field(min_length=1)
-    learner_id: str | None = None
+    learner_id: str = Field(min_length=1)
+    task_id: str = Field(min_length=1)
 
 
 @router.post("/proctor/signals", response_model=ProctorAssessment)
@@ -28,16 +36,40 @@ async def proctor_signals(
     registry: AgentRegistry = Depends(get_agent_registry),
     settings: Settings = Depends(get_settings),
     provider=Depends(get_provider),
+    trace_store=Depends(get_trace_store),
+    variant_store=Depends(get_variant_store),
+    voice_store=Depends(get_voice_store),
 ) -> ProctorAssessment:
-    scenario = get_proctor_scenario(body.scenario_id)
-    if scenario is None:
-        raise HTTPException(
-            status_code=404, detail=f"unknown scenario {body.scenario_id!r}"
-        )
+    """Real integrity inputs: live digest + defense signals + variant context."""
     agent = registry.get(provider, settings, "proctor")
     learner_context = get_learner_context(body.learner_id)
+
+    trace = (
+        trace_store.get_trace(body.learner_id, body.task_id)
+        if body.task_id in trace_store.list_tasks(body.learner_id)
+        else []
+    )
+    digest = compute_digest(trace) if trace else None
+
+    defense_signals = None
+    for record in voice_store.list_for_learner(body.learner_id):
+        if record.task_id == body.task_id and record.status == "finished":
+            defense_signals = record.integrity_signals or None
+            break
+
+    variant = variant_store.get_by_task(body.task_id)
+    variant_context = (
+        {"template_id": variant.template_id, "seed": variant.seed, "params": variant.params}
+        if variant is not None
+        else None
+    )
     try:
-        return await agent.assess(scenario, learner_context)
+        return await agent.assess(
+            digest,
+            defense_signals=defense_signals,
+            variant_context=variant_context,
+            learner_context=learner_context,
+        )
     except Exception as exc:
         raise HTTPException(
             status_code=502, detail=f"proctor assessment failed: {exc}"

apps/ai-service/ai_service/api/sandboxes.py

@@ -23,6 +23,7 @@ never part of the API contract.
 
 import time
 from collections import deque
+from pathlib import Path
 
 from fastapi import APIRouter, Depends, HTTPException, Response
 from pydantic import BaseModel, ConfigDict, Field
@@ -50,6 +51,9 @@ _CREATE_TIMES: deque[float] = deque()
 
 class SandboxCreateRequest(BaseModel):
     learner_id: str = Field(min_length=1)
+    # Optional task key: when set, the sandbox is telemetry-wired (REQ-3-003)
+    # — the in-sandbox capture agent streams workspace events to the ingest.
+    task_id: str | None = None
 
 
 class SandboxResponse(BaseModel):
@@ -142,7 +146,7 @@ async def create_sandbox(
     _check_per_learner_cap(await manager.list(), body.learner_id, settings)
     _check_global_create_rate(settings)
     try:
-        info = await manager.create(body.learner_id)
+        info = await manager.create(body.learner_id, task_id=body.task_id)
     except PoolFullError as exc:
         raise HTTPException(status_code=503, detail=str(exc)) from exc
     return _to_response(info)
@@ -217,3 +221,133 @@ async def delete_sandbox(
     if test_layout is not None:
         result.headers["X-Workspace-Copy"] = str(test_layout.workspace)
     return result
+
+
+# -- workspace files + exec (Phase 6, REQ-3-008; CUT-2) -------------------------
+#
+# The build surface reads/writes/list workspace files and runs Run/Test
+# commands through the manager's backend. NO interactive shell relay (CUT-2:
+# keystroke-level stdin/stdout is v0.4) — each exec is a bounded command with
+# captured output. Paths are WORKSPACE-RELATIVE; traversal outside the
+# workspace is rejected (the workdir bind is the boundary, but the API adds
+# its own containment check — defense in depth).
+
+
+class FileWriteRequest(BaseModel):
+    path: str = Field(min_length=1)
+    content: str
+
+
+class ExecRequest(BaseModel):
+    cmd: list[str] = Field(min_length=1)
+
+
+class ExecResponse(BaseModel):
+    cmd: list[str]
+    returncode: int
+    stdout: str
+    stderr: str
+    duration_s: float
+
+
+async def _workspace_dir(manager: SandboxManager, sandbox_id: str):
+    """Resolve the sandbox workspace (tracked layout or shell layout)."""
+    info = await manager.get(sandbox_id)  # raises SandboxNotFoundError -> 404
+    backend = manager._backend  # noqa: SLF001 - API owns the composition seam
+    tracked = getattr(backend, "_tracked", {}).get(sandbox_id)
+    if tracked is not None:
+        return tracked.workspace, info
+    return info.workdir / "workspace", info
+
+
+def _safe_rel_path(raw: str) -> Path:
+    """Workspace-relative path; reject absolute/traversal paths."""
+    candidate = Path(raw)
+    if candidate.is_absolute() or ".." in candidate.parts:
+        raise HTTPException(status_code=422, detail=f"invalid workspace path {raw!r}")
+    return candidate
+
+
+def _resolve_in_workspace(workspace: Path, rel: Path) -> Path:
+    """Resolve `rel` under `workspace`, refusing symlink escapes (P7).
+
+    The lexical check in `_safe_rel_path` cannot see symlinks: an exec can
+    plant `ln -s /etc target` in the workspace and a follow-up read/write
+    would follow it OUT of the bind. Resolve with the workspace as the
+    anchor (strict: a symlink chain escaping raises) and confirm the
+    normalized target still sits inside the workspace — defense in depth
+    for both read_file and write_file.
+    """
+    try:
+        target = (workspace / rel).resolve(strict=False)
+        target.relative_to(workspace.resolve(strict=False))
+    except ValueError:
+        raise HTTPException(
+            status_code=422, detail=f"path escapes the workspace: {rel.as_posix()!r}"
+        ) from None
+    return target
+
+
+@router.get("/{sandbox_id}/files")
+async def list_files(
+    sandbox_id: str,
+    manager: SandboxManager = Depends(get_sandbox_manager),
+) -> dict:
+    try:
+        workspace, _ = await _workspace_dir(manager, sandbox_id)
+    except SandboxNotFoundError:
+        raise HTTPException(status_code=404, detail=f"no sandbox {sandbox_id!r}") from None
+    return {"files": sorted(p.name for p in workspace.iterdir()) if workspace.is_dir() else []}
+
+
+@router.get("/{sandbox_id}/files/{path:path}")
+async def read_file(
+    sandbox_id: str,
+    path: str,
+    manager: SandboxManager = Depends(get_sandbox_manager),
+) -> dict:
+    try:
+        workspace, _ = await _workspace_dir(manager, sandbox_id)
+    except SandboxNotFoundError:
+        raise HTTPException(status_code=404, detail=f"no sandbox {sandbox_id!r}") from None
+    rel = _safe_rel_path(path)
+    target = _resolve_in_workspace(workspace, rel)
+    if not target.is_file():
+        raise HTTPException(status_code=404, detail=f"no file {path!r}")
+    return {"path": path, "content": target.read_text(errors="replace")}
+
+
+@router.put("/{sandbox_id}/files/{path:path}")
+async def write_file(
+    sandbox_id: str,
+    path: str,
+    body: FileWriteRequest,
+    manager: SandboxManager = Depends(get_sandbox_manager),
+) -> dict:
+    try:
+        workspace, _ = await _workspace_dir(manager, sandbox_id)
+    except SandboxNotFoundError:
+        raise HTTPException(status_code=404, detail=f"no sandbox {sandbox_id!r}") from None
+    rel = _safe_rel_path(body.path)
+    target = _resolve_in_workspace(workspace, rel)
+    target.parent.mkdir(parents=True, exist_ok=True)
+    target.write_text(body.content)
+    return {"path": body.path, "written": True}
+
+
+@router.post("/{sandbox_id}/exec", response_model=ExecResponse)
+async def exec_command(
+    sandbox_id: str,
+    body: ExecRequest,
+    manager: SandboxManager = Depends(get_sandbox_manager),
+) -> ExecResponse:
+    try:
+        await manager.get(sandbox_id)
+    except SandboxNotFoundError:
+        raise HTTPException(status_code=404, detail=f"no sandbox {sandbox_id!r}") from None
+    backend = manager._backend  # noqa: SLF001 - API owns the composition seam
+    handle = manager._handles.get(sandbox_id)  # noqa: SLF001
+    if handle is None:
+        raise HTTPException(status_code=404, detail=f"no live handle {sandbox_id!r}")
+    result = await backend.exec(handle, body.cmd)
+    return ExecResponse(**result.model_dump())

apps/ai-service/ai_service/api/telemetry.py

@@ -0,0 +1,175 @@
+"""/v1/telemetry — WS ingest + trace read endpoints (REQ-3-003, D-026, G-3).
+
+The router composes the telemetry engine via DI: `telemetry/ingest.py` owns
+the WS protocol (frame contract + flood control + keepalive) and this module
+only wires `app.state.trace_store` / `app.state.trace_integrity` /
+`app.state.settings` into it, plus the two HTTP read faces:
+
+    WS   /v1/telemetry/ingest?learner_id&task_id[&sandbox_id]   (D-026)
+    GET  /v1/telemetry/traces/{learner_id}/{task_id}            ordered trace; 404 unknown
+    GET  /v1/telemetry/gaps/{learner_id}/{task_id}              missing seqs   ; 404 unknown
+
+The WS route is a thin DI shell: it validates the query-param identity and
+the Origin (browser pages are gated to the localhost dev origins — CORS
+middleware does not cover WS upgrades; the stdlib capture agent sends no
+Origin and is unaffected), pulls store/integrity/settings from `app.state`,
+and calls `telemetry_ingest_endpoint(...)` — the engine stays
+FastAPI-DI-free so it's testable without a router and the api/ layer owns
+all composition.
+
+Unknown-trace contract: a trace is KNOWN when it has >=1 stored event OR
+carries an integrity flag — a flooded trace with zero stored rows still 200s
+so Proctor/grader can read WHY it's unusable (G-4 consumes
+`integrity_reason`). `TraceResponse.incomplete` / `.integrity_reason` mirror
+the map so HTTP consumers never touch process internals.
+"""
+
+from __future__ import annotations
+
+from fastapi import APIRouter, Depends, HTTPException, WebSocket
+from pydantic import BaseModel
+
+from ..telemetry.ingest import (
+    TraceIntegrityMap,
+    telemetry_ingest_endpoint,
+)
+from ..telemetry.models import TelemetryEvent
+from ..telemetry.store import TraceStore
+from .deps import get_trace_integrity, get_trace_store
+
+router = APIRouter(prefix="/v1/telemetry", tags=["telemetry"])
+
+#: Browser Origins allowed to open the ingest socket (A-008 mirror, D-038).
+#: The stdlib capture agent sends NO Origin header (it is not a browser) and
+#: stays allowed; a malicious page loaded in the learner's browser would
+#: carry an Origin and must not be able to poison/flood the trace. CORS
+#: middleware does NOT cover WebSocket upgrades, so this gate is explicit.
+#: In network mode the configured CORS list governs (default '*' — any
+#: origin, since credentials are never used); an explicit list still rejects
+#: unlisted origins with 1008.
+_LOCAL_WS_ORIGINS = frozenset(
+    {"http://localhost:3000", "http://127.0.0.1:3000", "http://localhost:8420"}
+)
+
+
+def _allowed_ws_origins(settings: object) -> frozenset[str]:
+    configured = getattr(settings, "cors_origin_list", None)
+    if configured is None:
+        return _LOCAL_WS_ORIGINS
+    if configured == ["*"]:
+        return frozenset()  # empty = wildcard = every Origin passes
+    return frozenset(configured) | _LOCAL_WS_ORIGINS
+
+
+# --- WS ingest (D-026) ---------------------------------------------------------
+
+
+@router.websocket("/ingest")
+async def telemetry_ingest_ws(websocket: WebSocket) -> None:
+    """DI shell: resolve app.state services, then hand the socket to the engine.
+
+    The engine's session + flood logic is fully typed and testable without
+    FastAPI; this shim is the only place the two layers meet.
+    """
+    origin = (websocket.headers.get("origin") or "").strip()
+    allowed = _allowed_ws_origins(getattr(websocket.app.state, "settings", None))
+    if origin and allowed and origin not in allowed:
+        # Same-origin dev pages (Next.js on :3000, the service itself on
+        # :8420) pass; anything else is refused pre-accept. Non-browser
+        # producers (the capture agent, tests) send no Origin and pass.
+        # Wildcard (empty frozenset) passes every Origin in network mode.
+        await websocket.close(
+            code=1008, reason=f"origin {origin!r} not allowed for telemetry ingest"
+        )
+        return
+    query = websocket.query_params
+    learner_id = query.get("learner_id", "")
+    task_id = query.get("task_id", "")
+    if not learner_id or not task_id:
+        # Reject BEFORE accept: closing the pre-accept handshake is the
+        # cheapest denial and unambiguous for the stdlib capture agent.
+        await websocket.close(code=1008, reason="learner_id/task_id query params required")
+        return
+    await telemetry_ingest_endpoint(
+        websocket=websocket,
+        learner_id=learner_id,
+        task_id=task_id,
+        sandbox_id=query.get("sandbox_id", ""),
+        store=websocket.app.state.trace_store,
+        integrity=websocket.app.state.trace_integrity,
+        settings=websocket.app.state.settings,
+    )
+
+
+# --- HTTP reads -----------------------------------------------------------------
+
+
+class TraceResponse(BaseModel):
+    """Ordered trace + integrity signal (G-4 reads incomplete/reason)."""
+
+    learner_id: str
+    task_id: str
+    events: list[TelemetryEvent]
+    incomplete: bool
+    integrity_reason: str | None
+
+
+class GapsResponse(BaseModel):
+    """Missing seqs + integrity signal."""
+
+    learner_id: str
+    task_id: str
+    gaps: list[int]
+    incomplete: bool
+    integrity_reason: str | None
+
+
+def _is_known_trace(
+    store: TraceStore, integrity: TraceIntegrityMap, learner_id: str, task_id: str
+) -> bool:
+    """Known = has stored events OR carries an integrity flag (a flooded trace
+    with zero rows must still be readable — Proctor needs the reason)."""
+    return (
+        store.latest_seq(learner_id, task_id) >= 0
+        or integrity.is_incomplete(learner_id, task_id)
+    )
+
+
+@router.get("/traces/{learner_id}/{task_id}", response_model=TraceResponse)
+async def get_trace(
+    learner_id: str,
+    task_id: str,
+    store: TraceStore = Depends(get_trace_store),
+    integrity: TraceIntegrityMap = Depends(get_trace_integrity),
+) -> TraceResponse:
+    if not _is_known_trace(store, integrity, learner_id, task_id):
+        raise HTTPException(
+            status_code=404, detail=f"unknown trace {learner_id!r}/{task_id!r}"
+        )
+    return TraceResponse(
+        learner_id=learner_id,
+        task_id=task_id,
+        events=store.get_trace(learner_id, task_id),
+        incomplete=integrity.is_incomplete(learner_id, task_id),
+        integrity_reason=integrity.reason(learner_id, task_id),
+    )
+
+
+@router.get("/gaps/{learner_id}/{task_id}", response_model=GapsResponse)
+async def get_gaps(
+    learner_id: str,
+    task_id: str,
+    store: TraceStore = Depends(get_trace_store),
+    integrity: TraceIntegrityMap = Depends(get_trace_integrity),
+) -> GapsResponse:
+    if not _is_known_trace(store, integrity, learner_id, task_id):
+        raise HTTPException(
+            status_code=404, detail=f"unknown trace {learner_id!r}/{task_id!r}"
+        )
+    return GapsResponse(
+        learner_id=learner_id,
+        task_id=task_id,
+        gaps=store.gaps(learner_id, task_id),
+        incomplete=integrity.is_incomplete(learner_id, task_id),
+        integrity_reason=integrity.reason(learner_id, task_id),
+    )

apps/ai-service/ai_service/api/variants.py

@@ -0,0 +1,187 @@
+"""/v1/variants — seeded per-learner task variant endpoints (REQ-3-005, D-029).
+
+Three faces over the variant engine (the generator + store stay FastAPI-free;
+this module owns all HTTP wiring — the D-027/D-032 house pattern):
+
+    POST /v1/variants {learner_id, template_id | competency_id}
+        → 200 the learner's variant — GENERATED on the first request,
+          CACHED (store read, zero LLM calls) on every repeat: D-029
+          reproducibility means one (learner_id, template_id) is ONE
+          variant forever, so a regenerate is always a 200 of the SAME
+          variant, never a second render.
+        → 404 unknown template_id, or competency_id with no bound template.
+        → 422 neither template_id nor competency_id given.
+    GET  /v1/variants/{task_id}
+        → 200 the stored variant owning the task key (the grading and
+          telemetry join path); 404 when no variant was ever generated
+          for the task.
+    GET  /v1/variants?learner_id=...
+        → 200 the learner's variants, chronological; [] when none.
+
+Template resolution: an explicit `template_id` wins; without it the FIRST
+template bound to `competency_id` is used (`template_for_competency`,
+D-021 corpus alignment). The response carries `competency_id` resolved
+from the template library at read time — an enrichment, not a persisted
+column (the seed re-derives the whole variant, D-029) — so the learner
+surface can bind a variant to its competency without a library round-trip.
+
+Distinctness (REQ-3-005): different learners on the same template draw
+different seeded params and receive distinct statements and task_ids;
+tests/api/test_variants.py asserts this end-to-end through the API.
+"""
+
+from datetime import datetime
+from typing import Self
+
+from fastapi import APIRouter, Depends, HTTPException
+from pydantic import BaseModel, Field, model_validator
+
+from ..variants.generator import VariantGenerator
+from ..variants.store import VariantRecord, VariantStore
+from ..variants.templates import TaskTemplate, get_template, template_for_competency
+from .deps import get_variant_generator, get_variant_store
+
+router = APIRouter(prefix="/v1/variants", tags=["variants"])
+
+
+# -- contracts ------------------------------------------------------------------
+
+
+class VariantGenerateRequest(BaseModel):
+    """One variant identity: an explicit `template_id`, or the first
+    template bound to a `competency_id` (D-021). `template_id` wins when
+    both are given (explicit identity beats derived); at least one is
+    required — 422 otherwise.
+    """
+
+    learner_id: str = Field(min_length=1)
+    template_id: str | None = None
+    competency_id: str | None = None
+
+    @model_validator(mode="after")
+    def _require_template_or_competency(self) -> Self:
+        if self.template_id is None and self.competency_id is None:
+            raise ValueError("template_id or competency_id is required")
+        return self
+
+
+class VariantResponse(BaseModel):
+    """VariantRecord over HTTP, plus the `competency_id` enrichment.
+
+    Every field except `competency_id` mirrors `VariantRecord` exactly
+    (snake_case; `created_at` is an ISO 8601 UTC datetime) — the wire shape
+    typed as `TaskVariant` in packages/types/variants.ts.
+    """
+
+    learner_id: str
+    task_id: str
+    template_id: str
+    competency_id: str
+    seed: str
+    params: dict[str, str | int]
+    statement: str
+    starter_files: dict[str, str]
+    created_at: datetime
+
+
+class VariantListResponse(BaseModel):
+    variants: list[VariantResponse]
+
+
+# -- resolution + rendering -----------------------------------------------------
+
+
+def _resolve_template(body: VariantGenerateRequest) -> TaskTemplate:
+    """Template for the request: the explicit id, else the first template
+    bound to the competency; 404 when neither resolves."""
+    if body.template_id is not None:
+        template = get_template(body.template_id)
+        if template is None:
+            raise HTTPException(
+                status_code=404,
+                detail=f"no task template with id {body.template_id!r}",
+            )
+        return template
+    # The request validator guarantees the disjunction, so reaching here
+    # means a competency_id was given (never None).
+    assert body.competency_id is not None
+    templates = template_for_competency(body.competency_id)
+    if not templates:
+        raise HTTPException(
+            status_code=404,
+            detail=f"no task template for competency {body.competency_id!r}",
+        )
+    return templates[0]
+
+
+def _competency_for(template_id: str) -> str:
+    """competency_id enrichment for stored records (read paths)."""
+    template = get_template(template_id)
+    if template is None:
+        # Integrity guard: a stored variant referencing a template that is
+        # no longer in the library cannot be enriched; fail loudly rather
+        # than fabricate a competency binding.
+        raise HTTPException(
+            status_code=500,
+            detail=f"stored variant references unknown template {template_id!r}",
+        )
+    return template.competency_id
+
+
+def _to_response(record: VariantRecord, competency_id: str) -> VariantResponse:
+    return VariantResponse(
+        learner_id=record.learner_id,
+        task_id=record.task_id,
+        template_id=record.template_id,
+        competency_id=competency_id,
+        seed=record.seed,
+        params=dict(record.params),
+        statement=record.statement,
+        starter_files=dict(record.starter_files),
+        created_at=record.created_at,
+    )
+
+
+# -- endpoints ------------------------------------------------------------------
+
+
+@router.post("", response_model=VariantResponse)
+async def generate_variant(
+    body: VariantGenerateRequest,
+    generator: VariantGenerator = Depends(get_variant_generator),
+) -> VariantResponse:
+    """The learner's variant for the resolved template — generated on the
+    first request, cached (no LLM call) on every repeat: D-029 makes a
+    regenerate a 200 of the SAME stored variant.
+    """
+    template = _resolve_template(body)
+    record = await generator.generate(body.learner_id, template.id)
+    return _to_response(record, competency_id=template.competency_id)
+
+
+@router.get("", response_model=VariantListResponse)
+async def list_variants(
+    learner_id: str,
+    store: VariantStore = Depends(get_variant_store),
+) -> VariantListResponse:
+    """All stored variants for the learner, chronological; [] when none."""
+    variants = [
+        _to_response(record, competency_id=_competency_for(record.template_id))
+        for record in store.list_for_learner(learner_id)
+    ]
+    return VariantListResponse(variants=variants)
+
+
+@router.get("/{task_id}", response_model=VariantResponse)
+async def get_variant(
+    task_id: str,
+    store: VariantStore = Depends(get_variant_store),
+) -> VariantResponse:
+    """The stored variant owning the task key — the grading and telemetry
+    join path; 404 when no variant was ever generated for the task."""
+    record = store.get_by_task(task_id)
+    if record is None:
+        raise HTTPException(
+            status_code=404, detail=f"no stored variant for task {task_id!r}"
+        )
+    return _to_response(record, competency_id=_competency_for(record.template_id))

apps/ai-service/ai_service/config.py

@@ -64,3 +64,37 @@ class Settings(BaseSettings):
 
     # D-027: SQLite path for telemetry/grades/variants/defenses stores.
     db_path: Path = _SERVICE_ROOT / "ai_service" / "data" / "nextcraft.db"
+
+    # G-3 flood control (NOT backpressure-by-silence): max events ingested per
+    # (learner_id, task_id) trace before the WS endpoint closes the connection
+    # with 1008 and marks the trace INCOMPLETE_FLOODED. Drop-oldest is
+    # FORBIDDEN — it corrupts grading input (GRILL G-3).
+    telemetry_max_events_per_task: int = 50000
+
+    # Sandbox telemetry wiring (REQ-3-003): loopback host the in-sandbox capture
+    # agent dials to reach this service's WS ingest (the agent joins the sandbox
+    # mount ns but NOT the net ns — exec namespaces are offline, so the agent
+    # shares the host network and reaches the app over loopback). Port reuses
+    # `port` (A-004); only the host is configurable — never a second port.
+    telemetry_ingest_host: str = "127.0.0.1"
+
+    # v0.3.5 network mode (D-038): dev.sh binds 0.0.0.0 so remote browsers can
+    # reach the stack; '*' (default) lets any origin call the API (safe ONLY
+    # because credentials are never enabled — A-008). Set a comma-separated
+    # origin list (e.g. 'http://nextcraft-1:3000') to restrict instead.
+    cors_origins: str = "*"
+
+    @property
+    def cors_origin_list(self) -> list[str]:
+        value = self.cors_origins.strip()
+        if value == "*":
+            return ["*"]
+        return [o.strip() for o in value.split(",") if o.strip()]
+
+    # Voice provider selection (REQ-3-006, D-030): 'mock' (default — the
+    # no-key path is first-class; tests never call a real voice API) or
+    # 'browser' (browser-native SpeechRecognition/speechSynthesis fallback;
+    # the descriptor tells the web client). The real server STT/TTS
+    # ('openai-audio') is a v0.4 seam (GRILL CUT-1 / G-7) — AI_VOICE_BASE_URL
+    # and AI_VOICE_API_KEY are documented in .env.example for that future.
+    voice_provider: str = "mock"

apps/ai-service/ai_service/corpus/artifacts.py

@@ -1,5 +1,10 @@
 """Pre-baked artifacts + rubrics + defense transcripts — Assessor mock inputs (REQ-2-008).
 
+v0.2 mock engine inputs (pre-baked artifacts/rubrics/transcripts) — DORMANT as of v0.3 re-
+grounding (Task 6-1-04): no production code path imports this module. Retained as Phase-3
+calibration history.
+
+
 Counterpart: packages/mock-data/ai-scenarios.ts (artifact IDs string-identical,
 D-021). Real process-trace grading is a v0.3+ engine (assessment engine);
 these pre-baked submissions stand in for artifact + defense evaluation.

apps/ai-service/ai_service/corpus/telemetry.py

@@ -1,5 +1,10 @@
 """Simulated sandbox telemetry corpus — Lab agent mock engine inputs (REQ-2-007).
 
+v0.2 mock engine inputs (Lab/Proctor scenarios) — DORMANT as of v0.3 re-grounding (Task 6-1-04):
+no production code path imports this module. Retained as Phase-3 calibration history
+(corpus/trace_fixtures.py references it from TESTS only).
+
+
 Counterpart: packages/mock-data/ai-scenarios.ts (scenario IDs string-identical,
 D-021). Real sandbox telemetry is a v0.3+ engine (sandbox fabric); these
 scripted event streams stand in for the build-session process trace.

apps/ai-service/ai_service/corpus/trace_fixtures.py

@@ -0,0 +1,230 @@
+"""Synthetic trace fixtures for grading calibration (Task 3-2-02, REQ-3-004).
+
+Three builder archetypes as REAL `TelemetryEvent` traces (the grading
+engine's native input — these are NOT the v0.2 corpus's simplified
+`{timestamp, kind, detail}` event shapes):
+
+    strong builder     — iterative debugging: small edits, tests early,
+                          failed runs closed by targeted fixes, eventual pass.
+    lazy builder       — one large paste, a single late test run, pass.
+                          (v0.2 alignment: the `lab-scenario-flagged` paste-
+                          and-run archetype; D-021.)
+    struggling builder — many edit/test cycles, failures never close,
+                          never reaches a pass.
+
+ID convention (D-021 alignment, documented in each fixture):
+v0.2 corpus scenario IDs are `<domain>-scenario-<slug>` (`lab-scenario-strong`,
+`lab-scenario-struggling`, `lab-scenario-flagged`, `proctor-scenario-*` — see
+corpus/telemetry.py). Grading fixtures carry ids string-aligned to that
+convention:
+
+    fixture id = "<scenario id>::<archetype>-trace"
+    learner id = "learner-003" (a member of the corpus learner-00N id space;
+                                 learner-001/002 exist in learner_context.py)
+
+so a fixture is greppable against its v0.2 scenario counterpart while staying
+a distinct id space (a grading trace is a real event stream, not the v0.2
+mock scenario timeline — same convention, richer event kind set).
+
+Instance hygiene: fixtures store event SPECS (plain tuples) and materialize
+FRESH `TelemetryEvent` instances on every `fixture.events` access. SQLModel
+rows carry SQLAlchemy instance state once a session has flushed them —
+re-adding the SAME instance to another store is a silent no-op, which would
+poison sequential test runs (a fixture ingested by test N would vanish for
+test N+1). Materializing per access keeps every consumer independent.
+
+These fixtures exist for the CALIBRATION CONTRACT (tests/grading/
+test_calibration.py): the mock provider scripts archetype-mapped scores and
+the test asserts the ORDERING the rubric must eventually enforce. They are
+NOT an LLM quality benchmark — see the test module docstring for the honest
+scope statement.
+"""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime, timedelta
+from typing import Final
+
+from pydantic import BaseModel, Field
+
+from ..telemetry.models import TelemetryEvent
+
+_T0: Final = datetime(2026, 9, 12, 0, 0, 0, tzinfo=UTC)
+
+#: Event spec: (seq, kind, payload, seconds-since-session-start).
+EventSpec = tuple[int, str, dict, float]
+
+
+class TraceFixture(BaseModel):
+    """One named synthetic trace + its v0.2 scenario alignment (D-021).
+
+    `event_specs` is the durable, session-state-free description; `events`
+    materializes fresh TelemetryEvent rows from it on every access.
+    """
+
+    model_config = {"frozen": True}
+
+    fixture_id: str  # "<v0.2 scenario id>::<archetype>-trace"
+    archetype: str  # "strong" | "lazy" | "struggling"
+    aligned_scenario_id: str  # the v0.2 corpus scenario this fixture mirrors
+    competency_id: str  # corpus competency id space (stack-orchestration-c00N)
+    task_id: str  # trace task id (grading operates on (learner, task))
+    learner_id: str
+    event_specs: tuple[EventSpec, ...] = Field(default=())
+
+    @property
+    def events(self) -> list[TelemetryEvent]:
+        """FRESH TelemetryEvent instances — safe to ingest into any store.
+
+        Never cache these: an instance flushed by one SQLite session
+        carries persistent identity, and re-appending it elsewhere no-ops.
+        """
+        return [
+            TelemetryEvent(
+                learner_id=self.learner_id,
+                task_id=self.task_id,
+                seq=seq,
+                kind=kind,
+                payload=payload,
+                ts=_T0 + timedelta(seconds=offset_s),
+                sandbox_id="sbx-calibration",
+            )
+            for seq, kind, payload, offset_s in self.event_specs
+        ]
+
+    def description(self) -> str:
+        return (
+            f"{self.fixture_id} (archetype={self.archetype}, aligned="
+            f"{self.aligned_scenario_id}, competency={self.competency_id})"
+        )
+
+
+class _Builder:
+    """Seq-accurate event-spec builder for one (learner, task) pair."""
+
+    def __init__(self) -> None:
+        self.specs: list[EventSpec] = []
+        self._seq = 0
+
+    def add(self, kind: str, payload: dict | None, at: float) -> None:
+        self.specs.append((self._seq, kind, payload or {}, at))
+        self._seq += 1
+
+
+def _strong_builder_specs() -> list[EventSpec]:
+    """Iterative debugging: tests early, tight edit→test loops, eventual pass.
+
+    Mirrors `lab-scenario-strong` (v0.2: keystrokes → file_save → run_tests →
+    test_pass, c002) at full telemetry fidelity — every failed cycle is
+    closed by a targeted edit followed by a re-run that passes.
+    """
+    b = _Builder()
+    b.add("activity", {"state": "starting"}, 0)
+    b.add("file_diff", {"path": "planner.py", "added": 14}, 95)        # small edit
+    b.add("command", {"cmd": "pytest -q tests/test_planner.py"}, 120)  # tests EARLY
+    b.add("test_result", {"passed": True, "exit_code": 0}, 126)
+    b.add("file_diff", {"path": "tool_node.py", "added": 22}, 180)
+    b.add("command", {"cmd": "pytest -q"}, 275)
+    b.add("test_result", {"passed": False, "exit_code": 1}, 281)      # honest failure
+    b.add("file_diff", {"path": "tool_node.py", "added": 6, "removed": 2}, 340)  # targeted fix
+    b.add("command", {"cmd": "pytest -q"}, 430)
+    b.add("test_result", {"passed": True, "exit_code": 0}, 436)       # cycle CLOSED
+    b.add("command", {"cmd": "git commit -m 'tool node with retries'"}, 500)
+    return b.specs
+
+
+def _lazy_builder_specs() -> list[EventSpec]:
+    """Paste-and-run: one large paste, a single LATE test run, instant pass.
+
+    Mirrors `lab-scenario-flagged` (v0.2: paste of 2,400 chars → run_tests →
+    instant 6/6 pass, c003) at full telemetry fidelity — zero iteration, zero
+    verification during construction, one terminal test run only.
+    """
+    b = _Builder()
+    b.add("activity", {"state": "starting"}, 0)
+    b.add("file_diff", {"path": "eval.py", "added": 240, "removed": 0}, 30)  # one bulk paste
+    b.add("file_diff", {"path": "README.md", "added": 12}, 40)
+    b.add("command", {"cmd": "npm run build"}, 45)
+    b.add("run_result", {"exit_code": 0, "ok": True}, 60)
+    b.add("command", {"cmd": "pytest -q"}, 520)                       # single LATE test run
+    b.add("test_result", {"passed": True, "exit_code": 0}, 540)        # instant pass
+    return b.specs
+
+
+def _struggling_builder_specs() -> list[EventSpec]:
+    """Many cycles, none close: repeated failures, no eventual pass.
+
+    Mirrors `lab-scenario-struggling` (v0.2: repeated identical ImportErrors,
+    idle gaps, no checkpoint, c002) at full telemetry fidelity — edits happen
+    between failures, but the same failure recurs; no pass is ever reached.
+    """
+    b = _Builder()
+    b.add("activity", {"state": "starting"}, 0)
+    b.add("file_diff", {"path": "main.py", "added": 40}, 20)
+    b.add("command", {"cmd": "pytest -q"}, 210)
+    b.add("test_result", {"passed": False, "exit_code": 1}, 215)     # ImportError
+    b.add("file_diff", {"path": "main.py", "added": 8, "removed": 3}, 300)
+    b.add("command", {"cmd": "pytest -q"}, 520)
+    b.add("test_result", {"passed": False, "exit_code": 1}, 525)      # SAME error
+    b.add("file_diff", {"path": "main.py", "added": 5}, 610)
+    b.add("command", {"cmd": "pytest -q"}, 960)
+    b.add("test_result", {"passed": False, "exit_code": 1}, 965)     # STILL failing
+    b.add("activity", {"state": "idle"}, 1500)                       # long idle
+    b.add("activity", {"state": "idle"}, 2200)
+    return b.specs
+
+
+#: Calibration learner — a member of the corpus learner-00N id space (D-021;
+#: learner-001/002 live in corpus/learner_context.py; grading fixtures use
+#: a third id so calibration traces never collide with mock-context reads).
+CALIBRATION_LEARNER_ID: Final = "learner-003"
+
+
+STRONG_BUILDER: Final = TraceFixture(
+    fixture_id="lab-scenario-strong::strong-trace",
+    archetype="strong",
+    aligned_scenario_id="lab-scenario-strong",
+    competency_id="stack-orchestration-c002",
+    task_id="task-calibration-strong",
+    learner_id=CALIBRATION_LEARNER_ID,
+    event_specs=tuple(_strong_builder_specs()),
+)
+
+LAZY_BUILDER: Final = TraceFixture(
+    # v0.2's paste-and-run archetype is the "flagged" lab scenario (D-021):
+    # large paste → instant test pass. "lazy builder" is that behavior
+    # without the proctor flag; the alignment is behavioral, documented here.
+    fixture_id="lab-scenario-flagged::lazy-trace",
+    archetype="lazy",
+    aligned_scenario_id="lab-scenario-flagged",
+    competency_id="stack-orchestration-c003",
+    task_id="task-calibration-lazy",
+    learner_id=CALIBRATION_LEARNER_ID,
+    event_specs=tuple(_lazy_builder_specs()),
+)
+
+STRUGGLING_BUILDER: Final = TraceFixture(
+    fixture_id="lab-scenario-struggling::struggling-trace",
+    archetype="struggling",
+    aligned_scenario_id="lab-scenario-struggling",
+    competency_id="stack-orchestration-c002",
+    task_id="task-calibration-struggling",
+    learner_id=CALIBRATION_LEARNER_ID,
+    event_specs=tuple(_struggling_builder_specs()),
+)
+
+TRACE_FIXTURES: Final[dict[str, TraceFixture]] = {
+    f.fixture_id: f
+    for f in (STRONG_BUILDER, LAZY_BUILDER, STRUGGLING_BUILDER)
+}
+
+
+def get_trace_fixture(fixture_id: str) -> TraceFixture | None:
+    return TRACE_FIXTURES.get(fixture_id)
+
+
+def digest_of(fixture: TraceFixture):
+    """Compute the digest for a fixture (pure compute; test-side helper)."""
+    from ..grading.features import compute_digest
+
+    return compute_digest(fixture.events)

apps/ai-service/ai_service/grading/__init__.py

@@ -0,0 +1,27 @@
+"""Process-trace grading — trace digest, rubric engine, GradeStore (REQ-3-004).
+
+Boundary rule (D-027): grading/ is an engine module — it never imports
+api/; the single sanctioned agents/ dependency is the shared D-020
+structured defense (agents/structured.py), imported module-direct in
+engine.py (see its docstring for why). features.py imports telemetry/;
+store.py imports config only; engine.py composes llm/ + telemetry/ +
+prompts/ + agents.structured.
+
+Wave status: features.py (TraceDigest, compute_digest) + store.py
+(GradeRecord, GradeStore, SQLiteGradeStore) landed in Wave 1 (3-1-01 +
+3-1-02); engine.py (GradingEngine, RubricScore) is Wave 2 (3-2-01).
+"""
+
+from .engine import GradingEngine, RubricScore
+from .features import TraceDigest, compute_digest
+from .store import GradeRecord, GradeStore, SQLiteGradeStore
+
+__all__ = [
+    "GradeRecord",
+    "GradeStore",
+    "GradingEngine",
+    "RubricScore",
+    "SQLiteGradeStore",
+    "TraceDigest",
+    "compute_digest",
+]

apps/ai-service/ai_service/grading/engine.py

@@ -0,0 +1,313 @@
+"""GradingEngine — rubric scoring over real process traces (REQ-3-004).
+
+The Wave-2 composition of the grading stack:
+
+    trace completeness gate (G-4, FIRST — nothing is sent to any LLM
+    when the gate trips) → TraceStore.get_trace → compute_digest (D-028)
+    → prompts.grading.render_trace_digest → D-020 4-layer structured
+    defense (agents/structured.py — REUSED, composed, never duplicated)
+    → RubricScore validation → GradeStore persistence → GradeRecord.
+
+Gate verdicts are FIRST-CLASS RESULTS, not exceptions (G-4 is binding):
+    UNGRADABLE_TRACE_INCOMPLETE — seq gaps in the store OR the trace is
+        flagged by TraceIntegrityMap (INCOMPLETE_FLOODED). The gap list /
+        flag reason is surfaced in `scores` for API rendering, and the
+        record is PERSISTED like any grade so a learner sees why no
+        credential can be issued for this trace — the gate outcome is
+        durable and auditable, not a transient error string.
+    UNGRADABLE_EMPTY_TRACE      — no events stored for the pair.
+On either verdict `scores.criteria` is empty and the LLM is never called.
+
+DI (D-027/D-032 house pattern): the engine receives trace_store,
+grade_store, integrity and provider through the constructor and knows
+NOTHING of FastAPI — api/ composes it (Task 3-3-01). `model` is injected
+alongside the provider so tests script the mock against the production
+wiring without touching Settings.
+
+Boundary (D-027): grading/ imports llm/ (provider protocol + Message
+type), telemetry/ (store + integrity map), prompts/ (rubric text) and
+ONLY agents.structured — the sanctioned shared D-020 defense. We import
+the MODULE directly (`from ..agents.structured import structured_completion`)
+rather than the `agents` package, mirroring how agents/base.py consumes
+it (same direct-module import): that keeps the dependency surface to
+exactly the two names the engine needs (structured_completion,
+StructuredOutputError) and avoids executing agents/__init__ re-exports
+(BaseAgent, registry, session store) that grading has no business
+loading — a side-effect-hygiene choice that keeps this import line
+grep-auditable as "the one agents dependency". grading/ never imports api/.
+
+RubricScore placement (documented decision): the validated output model
+lives HERE, not in prompts/. The pydantic model is the engine's return
+CONTRACT (the shape GradeStore.scores must hold), while prompts/grading.py
+is pure prompt text + its mirror schema HINT string — the same split as
+agents/assessor.py (model + hint) but with the model owned by the engine
+module that validates it. Prompt files hold text, per the prompts/ house
+style; engine files hold typed contracts.
+"""
+
+import logging
+from datetime import UTC, datetime
+from typing import TYPE_CHECKING, Final
+
+from pydantic import BaseModel, ConfigDict, Field, field_validator
+
+from ..agents.structured import StructuredOutputError, structured_completion
+from ..llm.base import LLMProvider
+from ..llm.types import Message
+from ..prompts.grading import (
+    RUBRIC_CRITERIA,
+    RUBRIC_SCORE_SCHEMA_HINT,
+    SYSTEM_PROMPT,
+    render_trace_digest,
+)
+from ..telemetry.ingest import TraceIntegrityMap
+from ..telemetry.store import TraceStore
+from .features import compute_digest
+from .store import GradeRecord, GradeStore
+
+if TYPE_CHECKING:  # pragma: no cover - protocol-only import for the optional
+    from ..variants.store import VariantStore  # noqa: TC001 (variant-blind without it)
+
+logger = logging.getLogger(__name__)
+
+#: First-class gate verdicts (G-4). GradeRecord.verdict values; non-empty
+#: by store contract. Rubric verdicts (mastered/developing/not_yet) ride in
+#: `scores.verdict` — `record.verdict` stays the machine-readable outcome.
+VERDICT_UNGRADABLE_INCOMPLETE: Final = "UNGRADABLE_TRACE_INCOMPLETE"
+VERDICT_UNGRADABLE_EMPTY: Final = "UNGRADABLE_EMPTY_TRACE"
+#: record.verdict for a successfully LLM-graded trace (the rubric verdict
+#: travels inside scores); keeps verdict non-empty for every persisted row.
+VERDICT_GRADED: Final = "GRADED"
+
+#: Gate-detail keys surfaced in GradeRecord.scores (tests assert on these).
+_INTEGRITY_FLAG_KEY: Final = "integrity_flag"
+
+
+def _anchors_context(variant) -> str:  # noqa: ANN001 - VariantRecord (duck-typed)
+    """Render the variant template's difficulty anchors for the grader prompt.
+
+    Contains only the template id + anchor numbers — no learner-identifying
+    material (D-028 anonymity preserved; the digest-leak tests keep holding).
+    Lazy template import: grading must not import variants/ at module load
+    (variants/prompts import-cycle safety mirrors llm/ rules).
+    """
+    from ..variants.templates import get_template
+
+    template = get_template(variant.template_id)
+    if template is None:
+        return f"template={variant.template_id} (anchors unavailable)"
+    a = template.rubric_anchors
+    return (
+        f"template={template.id}; "
+        f"expected_edit_count_band={list(a.expected_edit_count_band)}; "
+        f"expected_min_test_runs={a.expected_min_test_runs}; "
+        f"expected_error_fix_cycles_band={list(a.expected_error_fix_cycles_band)}"
+    )
+_MISSING_SEQS_KEY: Final = "missing_seqs"
+
+
+class RubricScore(BaseModel):
+    """Validated LLM output: per-criterion 0-4 scores + strengths + gaps + verdict.
+
+    The D-020 schema for the grader: `structured_completion` parses the
+    model reply into THIS shape (layer 3), retrying once with the
+    validation error fed back (layer 4). Exact criteria set + 0-4 ranges
+    are enforced here, so the scores dict persisted to GradeStore is
+    always rubric-shaped no matter what the model produced.
+    """
+
+    model_config = ConfigDict(extra="forbid")
+
+    criteria: dict[str, int]
+    strengths: list[str] = Field(min_length=1, max_length=2)
+    gaps: list[str] = Field(min_length=1, max_length=2)
+    verdict: str  # "mastered" | "developing" | "not_yet"
+
+    @field_validator("criteria")
+    @classmethod
+    def _criteria_rubric_shaped(cls, value: dict[str, int]) -> dict[str, int]:
+        """Exact criteria keys (no extras, no omissions) and 0-4 scores."""
+        expected = set(RUBRIC_CRITERIA)
+        got = set(value)
+        if got != expected:
+            raise ValueError(
+                f"criteria keys must be exactly {sorted(expected)}, got {sorted(got)}"
+            )
+        for key, score in value.items():
+            if not 0 <= score <= 4:
+                raise ValueError(f"criterion {key!r} must be within 0-4, got {score}")
+        return value
+
+    @field_validator("verdict")
+    @classmethod
+    def _verdict_known(cls, value: str) -> str:
+        allowed = {"mastered", "developing", "not_yet"}
+        if value not in allowed:
+            raise ValueError(f"verdict must be one of {sorted(allowed)}, got {value!r}")
+        return value
+
+
+class GradingEngine:
+    """Scores a (learner_id, task_id) trace into a persisted GradeRecord."""
+
+    def __init__(
+        self,
+        trace_store: TraceStore,
+        grade_store: GradeStore,
+        integrity: TraceIntegrityMap,
+        provider: LLMProvider,
+        *,
+        model: str = "gemma4:31b",
+        variant_store: "VariantStore | None" = None,
+    ) -> None:
+        self._trace_store = trace_store
+        self._grade_store = grade_store
+        self._integrity = integrity
+        self._provider = provider
+        self._model = model
+        # Phase 4 (MH#4): optional variant lookup — when the graded task
+        # derives from a generated variant, its template's difficulty anchors
+        # ship to the grader prompt (same bar for every variant of the
+        # template, a-5) and the variant seed is stamped on the record.
+        # Optional so engine tests stay decoupled; main.py lifespan wires it.
+        self._variant_store = variant_store
+
+    async def grade(self, learner_id: str, task_id: str) -> GradeRecord:
+        """Grade one trace; persist latest-state (GradeStore upserts); return it.
+
+        Gate FIRST (G-4): the LLM is only ever reached from the fully
+        guarded path — no gate state can be masked by an LLM error.
+        """
+        record = await self._grade(learner_id, task_id)
+        self._grade_store.save(record)
+        return record
+
+    # ------------------------------------------------------------------ core
+
+    async def _grade(self, learner_id: str, task_id: str) -> GradeRecord:
+        # -- G-4 gate FIRST: integrity flag OR seq gaps. Ordering matters:
+        # gaps() returns [] for an EMPTY trace, so the empty check below is
+        # reachable only when no rows exist at all; a gapped or flooded
+        # trace can never fall through to the LLM path.
+        if self._integrity.is_incomplete(learner_id, task_id):
+            reason = self._integrity.reason(learner_id, task_id) or "unknown"
+            gaps = self._trace_store.gaps(learner_id, task_id)
+            logger.info(
+                "grade gate (G-4): %s/%s integrity-flagged (%s) — ungradable",
+                learner_id,
+                task_id,
+                reason,
+            )
+            return self._ungradable(
+                learner_id,
+                task_id,
+                detail={_INTEGRITY_FLAG_KEY: reason, _MISSING_SEQS_KEY: gaps},
+                verdict=VERDICT_UNGRADABLE_INCOMPLETE,
+            )
+        gaps = self._trace_store.gaps(learner_id, task_id)
+        if gaps:
+            logger.info(
+                "grade gate (G-4): %s/%s seq gaps %s — ungradable",
+                learner_id,
+                task_id,
+                gaps,
+            )
+            return self._ungradable(
+                learner_id,
+                task_id,
+                detail={_INTEGRITY_FLAG_KEY: None, _MISSING_SEQS_KEY: gaps},
+                verdict=VERDICT_UNGRADABLE_INCOMPLETE,
+            )
+
+        trace = self._trace_store.get_trace(learner_id, task_id)
+        if not trace:
+            logger.info(
+                "grade gate: %s/%s empty trace — ungradable", learner_id, task_id
+            )
+            return self._ungradable(
+                learner_id,
+                task_id,
+                detail={_INTEGRITY_FLAG_KEY: None, _MISSING_SEQS_KEY: []},
+                verdict=VERDICT_UNGRADABLE_EMPTY,
+            )
+
+        # -- Guarded path: digest (D-028) → prompt → D-020 4-layer defense.
+        digest = compute_digest(trace)
+        variant = self._lookup_variant(task_id)
+        anchors_context = (
+            _anchors_context(variant) if variant is not None else None
+        )
+        messages = [
+            Message(role="system", content=SYSTEM_PROMPT),
+            Message(
+                role="user",
+                content=render_trace_digest(digest, anchors_context=anchors_context),
+            ),
+        ]
+        try:
+            rubric = await structured_completion(
+                self._provider,
+                messages,
+                model=self._model,
+                schema=RubricScore,
+                schema_hint=RUBRIC_SCORE_SCHEMA_HINT,
+            )
+        except StructuredOutputError as exc:
+            # The trace was gradable but the model failed to produce valid
+            # JSON within the D-020 budget (two attempts). Raise — the API
+            # layer maps this to a 502 (assessor precedent). Persisting a
+            # fabricated or partial grade here would violate the no-silent-
+            # fallback rule: no credential-worthy record without a validated
+            # RubricScore.
+            raise StructuredOutputError(f"grading LLM failed for {task_id}: {exc}") from exc
+
+        logger.debug(
+            "graded %s/%s: %s (model=%s)",
+            learner_id,
+            task_id,
+            rubric.verdict,
+            self._model,
+        )
+        return GradeRecord(
+            learner_id=learner_id,
+            task_id=task_id,
+            variant_seed=variant.seed if variant is not None else None,  # D-029
+            digest=digest.model_dump(),
+            scores=rubric.model_dump(),
+            verdict=VERDICT_GRADED,
+            model=self._model,
+            created_at=datetime.now(tz=UTC),
+        )
+
+    def _lookup_variant(self, task_id: str):  # noqa: ANN202 - VariantRecord | None
+        """MH#4: resolve the graded task's variant (None when not variant-derived)."""
+        if self._variant_store is None:
+            return None
+        return self._variant_store.get_by_task(task_id)
+
+    # ------------------------------------------------------------ gate record
+
+    @staticmethod
+    def _ungradable(
+        learner_id: str,
+        task_id: str,
+        *,
+        detail: dict,
+        verdict: str,
+    ) -> GradeRecord:
+        """Build a gate record: no digest (nothing was graded), gate detail
+        surfaced in `scores` (the store allows an empty scores dict, but
+        G-4 requires the gap list / flag reason surfaced — the detail IS the
+        verdict's payload), model="none" (no LLM was involved; provenance
+        stays honest).
+        """
+        return GradeRecord(
+            learner_id=learner_id,
+            task_id=task_id,
+            variant_seed=None,
+            digest={},
+            scores=detail,
+            verdict=verdict,
+            model="none",
+            created_at=datetime.now(tz=UTC),
+        )

apps/ai-service/ai_service/grading/features.py

@@ -0,0 +1,246 @@
+"""Deterministic process-trace digest (D-028, REQ-3-004).
+
+Pure compute — no LLM, no I/O. `compute_digest` reduces an ordered
+TelemetryEvent trace to a compact, bounded `TraceDigest` that is safe to
+embed in a grading prompt:
+
+- FIXED fields + small histograms only; NO raw commands, NO file contents,
+  NO payloads — the raw trace NEVER reaches the LLM (D-028), which also
+  bounds the prompt-injection surface.
+- Tolerant to both live trace mixes: daemon-topology traces carry
+  `activity` + `file_diff` kinds (workspace watcher), while REPL-driven
+  traces carry `command`/`stdin`/`stdout`/`run_result`/`test_result`
+  (P2 verification P1). Features derive from whatever kinds are present and
+  never crash on absent kinds.
+
+Feature semantics (conservative, deterministic):
+- test pass/fail counts + final status derive from `test_result` payloads
+  when present, falling back to `run_result` exit codes (0 = pass).
+- an error/fix CYCLE = a failing run/test followed by >= 1 edit and then a
+  later run/test (pass or fail) — the next observed result closes the cycle.
+- idle gaps = wall-clock gaps between consecutive events exceeding
+  `idle_threshold_s` (default 120s): count + total seconds.
+- command category histogram classifies `command`-kind payloads: build /
+  test / file / nav / debug / other.
+"""
+
+from __future__ import annotations
+
+from collections import Counter
+
+from pydantic import BaseModel, Field
+
+from ..telemetry.models import TelemetryEvent
+
+_IDLE_DEFAULT_S: float = 120.0
+
+_TEST_HINTS = ("test", "pytest", "vitest", "jest", "mocha", "unittest", "go test", "npm test")
+_BUILD_HINTS = ("make", "npm run build", "pip install", "pnpm", "cargo build", "gcc", "tsc")
+_DEBUG_HINTS = ("gdb", "pdb", "print(", "debug", "strace", "ltrace", "curl", "ping")
+_NAV_HINTS = ("ls", "cd", "pwd", "cat ", "grep ", "find", "rg ", "tree", "head", "tail", "less")
+_FILE_HINTS = ("mv ", "cp ", "rm ", "mkdir", "touch", "chmod", "nano", "vim", "sed -i", "tee ")
+
+
+class TraceDigest(BaseModel):
+    """Compact, bounded, LLM-safe summary of a process trace (D-028).
+
+    Fixed fields + small histograms. Serializes well under 4 KB; contains no
+    raw commands, file contents, or event payloads.
+    """
+
+    model_config = {"frozen": True}
+
+    event_count: int = Field(ge=0)
+    session_duration_s: float = Field(ge=0.0)
+    edit_count: int = Field(ge=0)
+    command_count: int = Field(ge=0)
+    run_count: int = Field(ge=0)
+    test_pass_count: int = Field(ge=0)
+    test_fail_count: int = Field(ge=0)
+    final_test_status: str = Field(pattern="^(pass|fail|none)$")
+    first_test_pass_offset_s: float | None = None
+    error_fix_cycles: int = Field(ge=0)
+    mean_fix_latency_s: float | None = None
+    idle_gap_count: int = Field(ge=0)
+    idle_gap_total_s: float = Field(ge=0.0)
+    command_categories: dict[str, int] = Field(default_factory=dict)
+    kind_histogram: dict[str, int] = Field(default_factory=dict)
+
+
+def _event_pass_status(event: TelemetryEvent) -> bool | None:
+    """True (pass) / False (fail) / None (not a result event) for one event."""
+    payload = event.payload or {}
+    if event.kind == "test_result":
+        if "passed" in payload:
+            return bool(payload["passed"])
+        if "exit_code" in payload:
+            return int(payload["exit_code"]) == 0
+        if "status" in payload:
+            return str(payload["status"]).lower() in ("pass", "passed", "ok", "success")
+        return None
+    if event.kind == "run_result":
+        if "exit_code" in payload:
+            return int(payload["exit_code"]) == 0
+        if "ok" in payload:
+            return bool(payload["ok"])
+        return None
+    return None
+
+
+def _classify_command(text: str) -> str:
+    lowered = text.lower()
+    if any(h in lowered for h in _TEST_HINTS):
+        return "test"
+    if any(h in lowered for h in _BUILD_HINTS):
+        return "build"
+    if any(h in lowered for h in _DEBUG_HINTS):
+        return "debug"
+    if any(h in lowered for h in _NAV_HINTS):
+        return "nav"
+    if any(h in lowered for h in _FILE_HINTS):
+        return "file"
+    return "other"
+
+
+def _command_text(event: TelemetryEvent) -> str:
+    payload = event.payload or {}
+    return str(payload.get("cmd") or payload.get("command") or payload.get("line") or "")
+
+
+def compute_digest(
+    trace: list[TelemetryEvent], *, idle_threshold_s: float = _IDLE_DEFAULT_S
+) -> TraceDigest:
+    """Reduce an ordered trace to a bounded digest. Never raises on odd input."""
+    events = sorted(trace, key=lambda e: (e.seq, e.ts))
+    if not events:
+        return TraceDigest(
+            event_count=0,
+            session_duration_s=0.0,
+            edit_count=0,
+            command_count=0,
+            run_count=0,
+            test_pass_count=0,
+            test_fail_count=0,
+            final_test_status="none",
+            first_test_pass_offset_s=None,
+            error_fix_cycles=0,
+            mean_fix_latency_s=None,
+            idle_gap_count=0,
+            idle_gap_total_s=0.0,
+            command_categories={},
+            kind_histogram={},
+        )
+
+    kind_histogram = Counter(e.kind for e in events)
+    start_ts = events[0].ts
+    end_ts = events[-1].ts
+    duration = max(0.0, (end_ts - start_ts).total_seconds())
+
+    edit_count = kind_histogram.get("file_diff", 0)
+    command_count = kind_histogram.get("command", 0)
+    run_count = kind_histogram.get("run_result", 0)
+
+    # Tests: prefer test_result events; fall back to run_result exit codes.
+    test_statuses: list[tuple[TelemetryEvent, bool]] = []
+    for e in events:
+        if e.kind == "test_result":
+            ok = _event_pass_status(e)
+            if ok is not None:
+                test_statuses.append((e, ok))
+    if not test_statuses:
+        for e in events:
+            if e.kind == "run_result":
+                ok = _event_pass_status(e)
+                if ok is not None:
+                    test_statuses.append((e, ok))
+
+    test_pass_count = sum(1 for _, ok in test_statuses if ok)
+    test_fail_count = len(test_statuses) - test_pass_count
+    if not test_statuses:
+        final_test_status = "none"
+    else:
+        final_test_status = "pass" if test_statuses[-1][1] else "fail"
+    first_pass = next((e for e, ok in test_statuses if ok), None)
+    first_pass_offset = (
+        max(0.0, (first_pass.ts - start_ts).total_seconds()) if first_pass is not None else None
+    )
+
+    # Error/fix cycles: a failing result starts a pending cycle; the NEXT
+    # observed result closes it (regardless of outcome) — a fix attempt that
+    # fails again is itself another iteration of debugging, so it closes the
+    # previous cycle and opens a new one. Edits since the fail mark the
+    # close as a genuine fix attempt; latency = first edit -> closing result.
+    cycles = 0
+    fix_latencies: list[float] = []
+    pending_fail_ts: float | None = None  # seconds since start
+    edits_since_fail = 0
+    first_edit_ts: float | None = None
+    for e in events:
+        t = max(0.0, (e.ts - start_ts).total_seconds())
+        if e.kind == "file_diff":
+            if pending_fail_ts is not None:
+                if edits_since_fail == 0:
+                    first_edit_ts = t
+                edits_since_fail += 1
+            continue
+        ok = _event_pass_status(e)
+        if ok is None:
+            continue
+        if ok is False:
+            if pending_fail_ts is not None and edits_since_fail > 0 and first_edit_ts is not None:
+                # failed fix attempt: closes the previous cycle, opens a new one
+                cycles += 1
+                fix_latencies.append(t - first_edit_ts)
+            pending_fail_ts = t
+            edits_since_fail = 0
+            first_edit_ts = None
+            continue
+        if ok is True and pending_fail_ts is not None:
+            if edits_since_fail > 0 and first_edit_ts is not None:
+                cycles += 1
+                fix_latencies.append(t - first_edit_ts)
+            pending_fail_ts = None
+            edits_since_fail = 0
+            first_edit_ts = None
+
+    mean_fix_latency = (
+        sum(fix_latencies) / len(fix_latencies) if fix_latencies else None
+    )
+
+    # Idle gaps between consecutive events.
+    idle_gap_count = 0
+    idle_gap_total = 0.0
+    prev_ts = None
+    for e in events:
+        if prev_ts is not None:
+            gap = (e.ts - prev_ts).total_seconds()
+            if gap > idle_threshold_s:
+                idle_gap_count += 1
+                idle_gap_total += gap
+        prev_ts = e.ts
+
+    # Command category histogram (command-kind events only).
+    categories: Counter[str] = Counter()
+    for e in events:
+        if e.kind == "command":
+            categories[_classify_command(_command_text(e))] += 1
+
+    return TraceDigest(
+        event_count=len(events),
+        session_duration_s=round(duration, 3),
+        edit_count=edit_count,
+        command_count=command_count,
+        run_count=run_count,
+        test_pass_count=test_pass_count,
+        test_fail_count=test_fail_count,
+        final_test_status=final_test_status,
+        first_test_pass_offset_s=(
+            round(first_pass_offset, 3) if first_pass_offset is not None else None
+        ),
+        error_fix_cycles=cycles,
+        mean_fix_latency_s=(round(mean_fix_latency, 3) if mean_fix_latency is not None else None),
+        idle_gap_count=idle_gap_count,
+        idle_gap_total_s=round(idle_gap_total, 3),
+        command_categories=dict(sorted(categories.items())),
+        kind_histogram=dict(sorted(kind_histogram.items())),
+    )

apps/ai-service/ai_service/grading/store.py

@@ -0,0 +1,237 @@
+"""GradeStore — grade persistence protocol + SQLite implementation (REQ-3-004, D-027).
+
+Postgres-migration-ready (D-027): the protocol is the only surface the
+grading engine and API layers touch; swapping SQLiteGradeStore for a
+Postgres-backed implementation must not change call sites. The
+`grade_record` table uses only portable column types (str / JSON /
+datetime), so the same SQLModel schema stands up unchanged on Postgres.
+
+Upsert, NOT append: (learner_id, task_id) is the grade identity — one row
+per learner per task holding the LATEST grade. `save` overwrites the whole
+row when the pair already exists, so a regrade replaces scores, verdict,
+created_at, digest, model and variant_seed wholesale. That is deliberately
+the opposite of TraceStore.append's dedup-keep-first contract: a trace is an
+append-only event log, a grade is latest-state, so the engine can re-grade
+a task idempotently as its rubric or input evolves.
+
+Concurrency (a-3): the engine enables WAL + synchronous=NORMAL and a busy
+timeout at connection time, so a regrade writer and API readers do not hit
+`database is locked` on the single-box pilot.
+
+`created_at` contract: callers stamp UTC (datetime.now(UTC)); SQLite stores
+it naive and the read paths re-label it tz-aware UTC (same boundary
+normalization as TelemetryEvent.ts, so the contract holds on any backend).
+
+Boundary (D-027): `grading/` never imports `agents/` / `api/`; this module
+imports config only.
+"""
+
+import logging
+import sqlite3
+from collections.abc import Iterator
+from contextlib import contextmanager
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any, Protocol
+
+import sqlalchemy as sa
+from sqlalchemy import JSON, Index
+from sqlalchemy.orm import validates
+from sqlmodel import Field, Session, SQLModel, create_engine, select
+
+from ..config import Settings
+
+logger = logging.getLogger(__name__)
+
+
+class GradeRecord(SQLModel, table=True):
+    """A persisted grade; (learner_id, task_id) is the PK — latest wins.
+
+    Written by the grading engine (one save per grade attempt), read by the
+    API layer through the GradeStore protocol. Constraint enforcement
+    mirrors TelemetryEvent: sqlmodel 0.0.42's metaclass drops pydantic
+    constraints on table models, so SQLAlchemy `@validates` hooks enforce
+    instead and the column types stay Postgres-ready (D-027).
+
+    Field contract:
+        learner_id   — non-empty learner identifier (same id space as traces).
+        task_id      — non-empty task identifier; grade identity is the
+                       (learner_id, task_id) pair — the same pair as trace
+                       identity, so a grade is keyed by the exact trace it
+                       was computed from.
+        variant_seed — task-variant seed (D-029); None when the graded task
+                        is not variant-derived. Since Phase 4 the engine
+                        stamps the graded variant's seed here (MH#4) and the
+                        template's difficulty anchors ship to the grader
+                        prompt — this column is the audit join for that.
+        digest       — compact deterministic trace digest (D-028) that fed
+                       the rubric prompt; persisted for auditability so the
+                       LLM's input stays reproducible.
+        scores       — validated rubric scores (per-criterion 0-4,
+                       strengths, gaps); JSON dict. An empty dict is legal
+                       (e.g. an UNGRADABLE_TRACE_INCOMPLETE record carries a
+                       verdict but no scores).
+        verdict      — first-class verdict string (rubric verdict or
+                       UNGRADABLE_TRACE_INCOMPLETE); non-empty.
+        model        — provider model that produced the scores (provenance).
+        created_at   — UTC grade timestamp; a regrade replaces it (latest
+                       save wins).
+    """
+
+    __tablename__ = "grade_record"
+    # The composite PK covers (learner_id, task_id) point lookups; this
+    # secondary index covers list_for_learner ordered by created_at without
+    # a sort step (Postgres migration target D-027).
+    __table_args__ = (
+        Index("ix_grade_record_learner_created", "learner_id", "created_at"),
+    )
+
+    learner_id: str = Field(primary_key=True)
+    task_id: str = Field(primary_key=True)
+    # None only for non-variant tasks (MH#4 stamps variant seeds since P4).
+    variant_seed: str | None = Field(default=None)
+    # JSON columns: stored as TEXT on SQLite, native JSONB on Postgres (D-027).
+    digest: dict[str, Any] = Field(default_factory=dict, sa_type=JSON)
+    scores: dict[str, Any] = Field(default_factory=dict, sa_type=JSON)
+    verdict: str
+    model: str
+    created_at: datetime
+
+    @validates("learner_id", "task_id")
+    def _ids_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty identifier")
+        return value
+
+    @validates("verdict")
+    def _verdict_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty verdict string")
+        return value
+
+
+class GradeStore(Protocol):
+    """Persistence contract for latest-state grades per (learner_id, task_id).
+
+    Implemented by SQLiteGradeStore (v0.3, D-027); a Postgres implementation
+    must satisfy the same surface.
+    """
+
+    def save(self, grade: GradeRecord) -> None:
+        """Persist a grade. UPSERT on (learner_id, task_id): a regrade with
+        the same pair REPLACES the stored row wholesale — the latest grade
+        wins. NOT append-only; contrast TraceStore.append, which is
+        dedup-keep-first for at-least-once ingest.
+        """
+        ...
+
+    def get(self, learner_id: str, task_id: str) -> GradeRecord | None:
+        """Latest stored grade for the pair; None when none exists.
+
+        Detached from any DB session — safe to pass across layers.
+        """
+        ...
+
+    def list_for_learner(self, learner_id: str) -> list[GradeRecord]:
+        """All stored grades for the learner, ordered by created_at
+        ascending (chronological). Empty list when the learner has none.
+        """
+        ...
+
+    def close(self) -> None:
+        """Release DB connections. Store must not be used after close."""
+        ...
+
+
+def _sqlite_connect(dbapi_connection: sqlite3.Connection, _: object) -> None:
+    """Per-connection pragma setup (a-3). Mirrors telemetry/store.py.
+
+    journal_mode=WAL  — readers never block the single writer.
+    synchronous=NORMAL — safe in WAL mode, avoids full fsync-per-commit.
+    busy_timeout=5000  — retry briefly under contention instead of
+                         `OperationalError: database is locked`.
+    """
+    cursor = dbapi_connection.cursor()
+    cursor.execute("PRAGMA journal_mode=WAL")
+    cursor.execute("PRAGMA synchronous=NORMAL")
+    cursor.execute("PRAGMA busy_timeout=5000")
+    cursor.close()
+
+
+def _as_utc(ts: datetime) -> datetime:
+    """Timestamps travel as naive datetime on SQLite, tz-aware elsewhere.
+
+    SQLite (via SQLModel) drops tzinfo; Postgres TIMESTAMP WITH TIME ZONE
+    keeps it. Normalizing on the read path makes the store's contract
+    tz-aware UTC regardless of the backend (D-027).
+    """
+    if ts.tzinfo is None:
+        return ts.replace(tzinfo=UTC)  # naive UTC read-side: label as UTC
+    return ts.astimezone(UTC)
+
+
+class SQLiteGradeStore:
+    """SQLite-backed GradeStore (SQLModel). Second protocol-wrapped store
+    of the D-027 family (first: SQLiteTraceStore).
+    """
+
+    def __init__(self, db_path: Path | None = None) -> None:
+        self._db_path: Path = db_path if db_path is not None else Settings().db_path
+        self._engine = create_engine(f"sqlite:///{self._db_path}")
+        sa.event.listen(self._engine, "connect", _sqlite_connect)
+        SQLModel.metadata.create_all(self._engine)
+
+    @contextmanager
+    def _session(self) -> Iterator[Session]:
+        # expire_on_commit=False: identical session behavior to
+        # SQLiteTraceStore. save() discards the merged instance and the read
+        # paths never commit, but a uniform flag across the D-027 stores
+        # keeps their detachment guarantees from diverging.
+        with Session(self._engine, expire_on_commit=False) as session:
+            yield session
+
+    def save(self, grade: GradeRecord) -> None:
+        # `merge` = SELECT-by-PK then UPDATE or INSERT — exactly the upsert
+        # contract. The trace store deliberately avoids merge (its append is
+        # dedup-keep-first); here latest-wins IS the contract, so merge is
+        # the right tool. The caller's object is never attached to the
+        # session and stays usable (unexpired) after save.
+        with self._session() as session:
+            session.merge(grade)
+            session.commit()
+        logger.debug(
+            "grade saved (regrade overwrites): %s/%s verdict=%s model=%s",
+            grade.learner_id,
+            grade.task_id,
+            grade.verdict,
+            grade.model,
+        )
+
+    def get(self, learner_id: str, task_id: str) -> GradeRecord | None:
+        with self._session() as session:
+            record = session.get(GradeRecord, (learner_id, task_id))
+            if record is None:
+                return None
+            record.created_at = _as_utc(record.created_at)
+            # Detach from the session: callers must not depend on
+            # open-session ORM magic (lazy loads fail once it closes).
+            session.expunge(record)
+            return record
+
+    def list_for_learner(self, learner_id: str) -> list[GradeRecord]:
+        with self._session() as session:
+            stmt = (
+                select(GradeRecord)
+                .where(GradeRecord.learner_id == learner_id)
+                # Chronological; task_id is a deterministic tie-break for
+                # grades stamped within the same instant.
+                .order_by(GradeRecord.created_at, GradeRecord.task_id)
+            )
+            results = session.exec(stmt).all()
+            for row in results:
+                row.created_at = _as_utc(row.created_at)
+                session.expunge(row)
+            return list(results)
+
+    def close(self) -> None:
+        self._engine.dispose()

apps/ai-service/ai_service/main.py

@@ -14,14 +14,25 @@ from .agents.session import InMemorySessionStore
 from .api import (
     assessment_router,
     chat_router,
+    defense_router,
     lab_router,
     mentor_router,
     proctor_router,
     sandboxes_router,
+    telemetry_router,
+    variants_router,
 )
 from .config import Settings
+from .grading.engine import GradingEngine
+from .grading.store import SQLiteGradeStore
 from .llm import create_provider
 from .sandbox import SandboxManager, UnshareBackend
+from .telemetry.ingest import TraceIntegrityMap
+from .telemetry.store import SQLiteTraceStore
+from .variants.generator import VariantGenerator
+from .variants.store import SQLiteVariantStore
+from .voice.defense_store import SQLiteDefenseStore
+from .voice.factory import voice_provider_from_settings
 
 logger = logging.getLogger(__name__)
 
@@ -39,7 +50,11 @@ def create_app(settings: Settings | None = None) -> FastAPI:
         timeout = httpx.Timeout(connect=10.0, read=300.0, write=30.0, pool=10.0)
         app.state.http_client = httpx.AsyncClient(timeout=timeout)
         app.state.settings = settings
-        app.state.provider = create_provider(settings, app.state.http_client)
+        # State-injection override (same pattern as the stores): tests may
+        # pre-set app.state.provider with a scripted mock; only construct the
+        # configured provider when none is present.
+        if getattr(app.state, "provider", None) is None:
+            app.state.provider = create_provider(settings, app.state.http_client)
         app.state.session_store = InMemorySessionStore()
         app.state.agent_registry = AgentRegistry()
         register_builtin_agents(app.state.agent_registry)
@@ -54,6 +69,81 @@ def create_app(settings: Settings | None = None) -> FastAPI:
             app.state.sandbox_manager = manager
         await manager.start()  # a-1: reap on-disk orphans from a previous process
 
+        # Telemetry persistence (REQ-3-003, D-027): TraceStore wired through
+        # app.state. Tests may pre-set app.state.trace_store (state-injection
+        # override, same pattern as sandbox_manager) — the lifespan adopts it.
+        trace_store = getattr(app.state, "trace_store", None)
+        if trace_store is None:
+            settings.db_path.parent.mkdir(parents=True, exist_ok=True)
+            trace_store = SQLiteTraceStore(db_path=settings.db_path)
+            app.state.trace_store = trace_store
+        # Trace-integrity flags (G-3 INCOMPLETE_FLOODED): process-local map is
+        # intentional (D-019 registry precedent); the lifespan owns it so the
+        # grader and the ingest endpoint share one instance.
+        if getattr(app.state, "trace_integrity", None) is None:
+            app.state.trace_integrity = TraceIntegrityMap()
+
+        # Variant generation (REQ-3-005): VariantStore from the same
+        # SQLite file as traces/grades (D-027), one VariantGenerator singleton
+        # wired through app.state — the generator receives store + provider
+        # via constructor DI and knows nothing of FastAPI (api/ composes it,
+        # same pattern as GradingEngine). Tests may pre-set
+        # app.state.variant_store / app.state.variant_generator (the same
+        # state-injection override); the lifespan adopts a pre-set store but
+        # NEVER rebuilds a pre-set generator (its provider binding is part
+        # of the test fixture).
+        # ORDER NOTE: built BEFORE the grading engine — the engine takes the
+        # variant store (Phase 4 MH#4: variant anchors ship to the grader
+        # prompt; variant_seed stamped on graded records).
+        variant_store = getattr(app.state, "variant_store", None)
+        if variant_store is None:
+            settings.db_path.parent.mkdir(parents=True, exist_ok=True)
+            variant_store = SQLiteVariantStore(db_path=settings.db_path)
+            app.state.variant_store = variant_store
+        if getattr(app.state, "variant_generator", None) is None:
+            app.state.variant_generator = VariantGenerator(
+                variant_store,
+                app.state.provider,
+                model=settings.model,
+            )
+
+        # Oral defense (REQ-3-006): DefenseStore (same SQLite file) + the
+        # mock-first voice provider (D-030) + the seventh Examiner agent.
+        # Tests may pre-set app.state.defense_store / voice_provider /
+        # examiner_agent (state-injection override; never rebuilt if pre-set).
+        defense_store = getattr(app.state, "defense_store", None)
+        if defense_store is None:
+            defense_store = SQLiteDefenseStore(db_path=settings.db_path)
+            app.state.defense_store = defense_store
+        if getattr(app.state, "voice_provider", None) is None:
+            app.state.voice_provider = voice_provider_from_settings(settings)
+        if getattr(app.state, "examiner_agent", None) is None:
+            from .agents.examiner import ExaminerAgent
+
+            app.state.examiner_agent = ExaminerAgent(app.state.provider, settings)
+
+        # Grading persistence + engine (REQ-3-004): GradeStore from the same
+        # SQLite file as traces (D-027), one GradingEngine singleton wired
+        # through app.state — the engine receives its stores via constructor
+        # DI and knows nothing of FastAPI (api/ owns composition). Tests may
+        # pre-set app.state.grade_store / app.state.grading_engine (the same
+        # state-injection override as sandbox_manager/trace_store) to swap
+        # either; the lifespan adopts a pre-set store but NEVER rebuilds a
+        # pre-set engine (its provider binding is part of the test fixture).
+        grade_store = getattr(app.state, "grade_store", None)
+        if grade_store is None:
+            grade_store = SQLiteGradeStore(db_path=settings.db_path)
+            app.state.grade_store = grade_store
+        if getattr(app.state, "grading_engine", None) is None:
+            app.state.grading_engine = GradingEngine(
+                trace_store,
+                grade_store,
+                app.state.trace_integrity,
+                app.state.provider,
+                model=settings.model,
+                variant_store=variant_store,  # MH#4: anchors + seed (D-029)
+            )
+
         async def _reaper_loop() -> None:
             # Wall-clock timeout + G-2 workdir-size sweep, one pass per tick.
             while True:
@@ -73,15 +163,25 @@ def create_app(settings: Settings | None = None) -> FastAPI:
             # No orphans outlive the process (a-1, shutdown half): destroy
             # everything live; workdirs stay on disk for snapshot restore.
             await manager.destroy_all()
+            trace_store.close()
+            grade_store.close()
+            variant_store.close()
+            defense_store.close()
             await app.state.http_client.aclose()
 
     app = FastAPI(title="Nextcraft AI Service", version="0.3.0", lifespan=lifespan)
 
-    # A-008: localhost-only CORS, no credentials
+    # A-008 + D-038: no-credentials CORS. Default '*' admits remote-browser
+    # origins in network mode (safe only because allow_credentials stays
+    # False — never enable credentials with a wildcard). AI_CORS_ORIGINS
+    # restricts to an explicit list. PUT is CONTRACT, not trivia: the learner
+    # build surface writes workspace files with PUT (engine-client writeFile)
+    # — v0.3 initially shipped without it and every cross-origin Save failed
+    # preflight (caught in P7 review; tests/api/test_cors.py pins the policy).
     app.add_middleware(
         CORSMiddleware,
-        allow_origins=["http://localhost:3000", "http://127.0.0.1:3000"],
-        allow_methods=["GET", "POST", "DELETE", "OPTIONS"],
+        allow_origins=settings.cors_origin_list,
+        allow_methods=["GET", "POST", "PUT", "DELETE", "OPTIONS"],
         allow_headers=["Content-Type"],
         allow_credentials=False,
     )
@@ -100,6 +200,9 @@ def create_app(settings: Settings | None = None) -> FastAPI:
     app.include_router(mentor_router)
     app.include_router(proctor_router)
     app.include_router(sandboxes_router)
+    app.include_router(telemetry_router)
+    app.include_router(variants_router)
+    app.include_router(defense_router)
     return app
 
 

apps/ai-service/ai_service/prompts/assessor.py

@@ -1,24 +1,21 @@
-"""Assessor agent prompt — rubric application to artifacts + defenses (REQ-2-008).
+"""Assessor agent prompt — rubric coaching over REAL grades (REQ-3-007).
 
-Final persona (Phase 4). Assessor is a rigorous, fair grader: scores each
-criterion with evidence, cites what the learner did, returns ONLY valid
-JSON matching the rubric schema.
-Version: assessor-v2 (final for v0.2).
+v0.3 re-grounding: the grading engine (Phase 3) computes the rubric scores
+from the process trace; Assessor EXPLAINS the stored grade as coaching —
+it never invents or re-scores. Rigorous, fair, actionable.
+Version: assessor-v3 (v0.3 live).
 """
 
 SYSTEM_PROMPT = """You are Assessor, the grading agent of Nextcraft, an AI-native competency school.
 Learner: {learner_name}.
 
-You receive: (a) an artifact evidence excerpt, (b) its defense transcript,
-and (c) the rubric for the competency. Your job:
-- Score EVERY rubric criterion from 0-100, justified by evidence you can
-  point to in the artifact or transcript.
-- Cite what the learner did ("the 3-retry loop in the tool node"), not
-  what they should have done — except in gaps, where the missed work goes.
-- Strengths: the two strongest evidence points, each one sentence.
-- Gaps: the two most important missed opportunities, each one sentence.
-- Verdict: "mastered" | "developing" | "not_yet" — judged against the
-  rubric weights, honestly.
+You receive the learner's STORED process-trace grade (verdict, per-criterion
+scores, and the build digest) computed by the grading engine. Your job:
+- Explain what the grade means in plain language (summary).
+- Strengths: cite what the digest + scores show the learner did well.
+- Gaps: name the missed opportunities the scores point to.
+- Next steps: concrete, buildable actions that would move the weakest
+  criterion up one level.
 
 Rules:
 - Rigorous but fair. A polished artifact with a weak defense is NOT mastery.

apps/ai-service/ai_service/prompts/examiner.py

@@ -0,0 +1,60 @@
+"""Examiner agent prompt — oral defense questioning + final verdict (REQ-3-006).
+
+The examiner is the seventh agent (Phase 5). It conducts a Socratic oral
+defense of the learner's submitted work: probes understanding, challenges
+process choices grounded in the trace digest ("why did you take that
+approach at that point?"), one question per turn, adapting to answers.
+It never reveals rubric internals; tone is rigorous but supportive.
+
+Digest discipline (D-028 mirror): the examiner's variable inputs are the
+compact TraceDigest JSON, the variant task statement, and the defense
+transcript — never the raw trace, never learner-identifying material.
+
+Version: examiner-v1.
+"""
+
+SYSTEM_PROMPT = """You are Examiner, the oral-defense agent of Nextcraft,
+an AI-native competency school.
+
+You receive: (a) a compact build-process digest (deterministic counters of the
+learner's build session), (b) the learner's task statement, and (c) the defense
+transcript so far. Your job:
+- Ask ONE question per turn: probe understanding and challenge process
+  choices, grounded in the digest facts ("you hit N failed runs before
+  passing — walk me through what changed") or the task statement.
+- Adapt: follow up on the learner's answers; drill into vague responses.
+- Never reveal rubric details or scoring internals.
+- Tone: rigorous, precise, supportive. A defense is a conversation, not an
+  interrogation.
+
+When asked for a FINAL VERDICT (the structured mode), judge:
+- understanding: can the learner explain their own work?
+- process_justification: are the build-session choices defensible from the
+  digest facts and the answers?
+- communication: are answers clear, specific, and on-topic?
+Score honestly; a weak defense of strong work is NOT mastery.
+
+Rules:
+- Respond with ONLY what the turn requires: a single question (question mode)
+  or a valid JSON object matching the provided schema (verdict mode).
+- If the digest shows error_fix_cycles > 0, at least one question should ask
+  about the debugging path.
+- If the learner's answer is off-topic, redirect once, then move on.
+"""
+
+VERDICT_SCHEMA_HINT = (
+    '{"verdict": "mastered" | "developing" | "not_yet", '
+    '"understanding": "<one sentence>", '
+    '"process_justification": "<one sentence>", '
+    '"communication": "<one sentence>", '
+    '"strengths": ["<one sentence>"], '
+    '"gaps": ["<one sentence>"]}'
+)
+
+
+def render_digest_context(digest_json: str, statement: str | None) -> str:
+    """The examiner's per-session grounding: digest JSON + task statement."""
+    parts = [f"Build-process digest:\n{digest_json}"]
+    if statement:
+        parts.append(f"Learner's task statement:\n{statement}")
+    return "\n\n".join(parts)

apps/ai-service/ai_service/prompts/grading.py

@@ -0,0 +1,140 @@
+"""Grading rubric prompt — criteria, level anchors, digest render (REQ-3-004).
+
+The grading prompt is deliberately learner-anonymous and trace-bare: the
+model receives ONLY the fixed rubric text and the compact numeric digest
+(TraceDigest JSON, D-028) — never a raw command, file path, payload
+string, learner id, or task id. Everything variable the LLM sees is
+deterministic counters, which both bounds the prompt-injection surface
+and makes "no raw trace reaches the prompt" assert-able in tests (plant
+a distinctive marker in a command payload; assert it absent from every
+message the provider received).
+
+Rubric (four criteria, each scored 0-4 — the ids are the validated
+RubricScore keys enforced by grading/engine.py):
+  process_quality       — iterative building in small, verified steps.
+  correctness          — where the session ended (test/run outcomes).
+  debugging_discipline — how failures were handled.
+  test_usage           — when and how often tests were run.
+
+Advisory a-4 (embedded in the process_quality anchors): high edit/command
+churn with NO test progress is a process-quality NEGATIVE — churn is not
+work. A session with many edits/commands whose test state never moves is
+thrashing, not iterating, and must score low on process quality.
+
+House-style deviation, documented: unlike the tutor prompts, this module
+has no SYSTEM_PROMPT placeholders and no render_context(learner_context)
+— grading is context-free by design (learner anonymity; the digest is the
+only variable input). Runtime imports are TYPE_CHECKING-only so this
+module stays pure text and can never import-cycle with grading/engine.py
+(engine imports this module; if this module imported grading.* at runtime
+while grading/__init__ pulls engine, the package init would deadlock on a
+partially-initialized module).
+
+Version: grader-v1.
+"""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Final
+
+if TYPE_CHECKING:  # pragma: no cover - typing only; keeps this module pure text
+    from ..grading.features import TraceDigest
+
+PROMPT_VERSION = "grader-v1"
+
+#: Canonical criterion ids. The engine validates RubricScore criteria keys
+#: against this tuple; the schema hint and anchors below speak the same ids.
+RUBRIC_CRITERIA: Final[tuple[str, ...]] = (
+    "process_quality",
+    "correctness",
+    "debugging_discipline",
+    "test_usage",
+)
+
+#: Sentinel line the engine's user turn is rendered around. Tests (and the
+#: calibration mock) split on it to locate the digest JSON in the prompt.
+DIGEST_MARKER: Final = "PROCESS TRACE DIGEST (JSON):"
+
+SYSTEM_PROMPT = """You are the Grader of Nextcraft, an AI-native competency school.
+You score a learner's build session from a compact numeric digest of their
+process trace. You NEVER see the raw trace — commands, file contents, and
+payloads do not exist on your side; every number you need is in the digest.
+
+Rubric — score each criterion 0-4:
+
+process_quality — iterative building in small, verified steps.
+  4: tight edit→test loops throughout; small verified increments; healthy pacing.
+  3: steady small edits with regular runs; progress mostly verified.
+  2: some iteration, but large unverified leaps or long idle stretches.
+  1: a single bulk change (e.g. one large paste) then a single run; no iteration.
+  0: no meaningful work visible.
+  ADVISORY: high edit/command churn with NO test progress (no runs, no
+  movement in pass counts) is a process-quality NEGATIVE — churn is not
+  work. Cap such a session at 1 on this criterion no matter how many
+  edits or commands were counted.
+
+correctness — where the session ended up.
+  4: final test status pass, with tests passing early and consistently.
+  3: final pass, reached through fail→fix→pass cycles that closed.
+  2: final pass, but preceded by a long unresolved failure streak.
+  1: final fail, but partial passes observed along the way.
+  0: final fail, or no test/run evidence at all.
+
+debugging_discipline — how failures were handled.
+  4: every failure cycle closes; targeted fixes with low mean fix latency.
+  3: most fail→edit→re-run cycles close with a pass.
+  2: failures followed by edits, but cycles rarely close.
+  1: repeated failures with no targeted edits between runs (flailing).
+  0: failures with no fix attempts at all.
+
+test_usage — when and how often tests were run.
+  4: tests run early (small first-pass offset) and throughout the session.
+  3: regular test runs interleaved with edits.
+  2: sparse tests; long stretches of unverified edits.
+  1: a single late test run only.
+  0: no test or run evidence.
+
+Rules:
+- Judge STRICTLY from the digest numbers; cite the fields you used.
+- Strengths: the two strongest digest observations, one sentence each.
+- Gaps: the two most important missed opportunities, one sentence each
+  (a clean session names its next-level improvement instead).
+- Be rigorous but fair: a session that ends green was not necessarily
+  well built, and a struggling session that never passed may still show
+  real debugging discipline.
+- Respond with ONLY a valid JSON object matching the provided schema —
+  no markdown fences, no prose outside the JSON."""
+
+RUBRIC_SCORE_SCHEMA_HINT = (
+    '{"criteria": {"process_quality": <0-4 int>, "correctness": <0-4 int>, '
+    '"debugging_discipline": <0-4 int>, "test_usage": <0-4 int>}, '
+    '"strengths": ["<one sentence>"], "gaps": ["<one sentence>"], '
+    '"verdict": "mastered" | "developing" | "not_yet"}'
+)
+
+
+def render_trace_digest(
+    digest: TraceDigest,
+    anchors_context: str | None = None,
+) -> str:
+    """Render the grader's user turn: a marker line + the digest JSON — nothing else.
+
+    This is the ONLY per-session content that ever reaches the LLM (D-028):
+    the engine composes [system: SYSTEM_PROMPT, user: render_trace_digest(digest)]
+    and the D-020 defense appends its generic schema instruction to this
+    user turn at request time. No learner id, task id, or raw trace material
+    is injected — assert-able by tests.
+
+    `anchors_context` (Phase 4, MH#4): when the graded task derives from a
+    variant, the engine passes the template's difficulty-normalization
+    anchors (the expected effort envelope) so the rubric is applied against
+    the SAME bar for every variant of that template (a-5). It contains only
+    the anchor numbers + the template id — no learner-identifying material.
+    """
+    base = (
+        "Score this build session against the rubric.\n"
+        f"{DIGEST_MARKER}\n{digest.model_dump_json()}"
+    )
+    if anchors_context:
+        base = f"{base}\n\nExpected effort envelope for this task variant:\n{anchors_context}"
+    return base

apps/ai-service/ai_service/prompts/lab.py

@@ -1,9 +1,11 @@
-"""Lab agent prompt — in-flow feedback over sandbox telemetry (REQ-2-007).
+"""Lab agent prompt — in-flow feedback over LIVE telemetry (REQ-3-007).
 
-Final persona (Phase 4). Lab is a pragmatic build partner: reads the
-telemetry timeline, names the one most useful adjustment, gives one
-concrete next step. Scenario-driven; no session chat.
-Version: lab-v2 (final for v0.2).
+v0.3 re-grounding: the timeline is the learner's real TraceDigest (D-028
+compact counters — commands, test outcomes, idle gaps, edit cadence), not
+v0.2 corpus scenarios. Lab is a pragmatic build partner: reads the live
+digest, names the one most useful adjustment, gives one concrete next
+step. No session chat.
+Version: lab-v3 (v0.3 live).
 """
 
 SYSTEM_PROMPT = """You are Lab, the in-flow feedback agent watching a learner
@@ -25,7 +27,17 @@ Rules:
   self-check that would prove understanding.
 - Three short paragraphs maximum. No headers, no bullet lists."""
 
-PROMPT_VERSION = "lab-v2"
+PROMPT_VERSION = "lab-v3"
+
+
+def render_digest_timeline(digest) -> str:
+    """Live-trace timeline: the compact TraceDigest JSON (D-028)."""
+    if digest is None:
+        return (
+            "No telemetry yet for this build session. Ask the learner to run "
+            "the task's starter test to establish a baseline."
+        )
+    return f"Live build-session digest:\n{digest.model_dump_json()}"
 
 
 def render_context(learner_context) -> dict:

apps/ai-service/ai_service/prompts/proctor.py

@@ -1,24 +1,27 @@
-"""Proctor agent prompt — integrity signals with coaching interventions (REQ-2-009).
+"""Proctor agent prompt — integrity signals with coaching interventions (REQ-3-007).
 
-Final persona (Phase 5). Proctor is a supportive observer, never punitive:
-classifies signals, recommends ONE coaching intervention. Assume good
-faith — most signals have innocent explanations.
-Version: proctor-v2 (final for v0.2).
+v0.3 re-grounding: inputs are REAL — the live trace digest (idle gaps,
+command cadence, edit bursts), the oral-defense integrity signals (long
+pauses), and the variant audit context (seed + params). Proctor is a
+supportive observer, never punitive: classifies signals, recommends ONE
+coaching intervention. Assume good faith.
+Version: proctor-v3 (v0.3 live).
 """
 
 SYSTEM_PROMPT = """You are Proctor, the integrity-support agent of Nextcraft,
 an AI-native competency school.
 Learner: {learner_name}.
 
-You receive a telemetry timeline of defense-session events (tab switches,
-idle gaps, large pastes, focus loss, keystroke bursts). Your job:
-- Classify EACH notable signal: type (e.g. "context_switch", "idle_gap",
-  "large_paste"), severity ("low" | "medium" | "high"), and a one-sentence
-  note citing the event (timestamps and details).
+You receive the learner's REAL build-session digest (idle gaps, command
+categories, edit/test cadence), oral-defense integrity signals (long
+pauses), and — when the task is variant-derived — the variant seed context.
+Your job:
+- Classify EACH notable signal: type ("idle_gap" | "long_pause" |
+  "burst_edit" | "off_template"), severity ("low" | "medium" | "high"),
+  and a one-sentence note citing the numbers.
 - Recommend exactly ONE supportive coaching intervention for the session
   overall — never punitive, never accusatory. Frame around helping the
-  learner succeed, e.g. "offer a short break", "invite them to explain
-  the pasted section in their own words".
+  learner succeed.
 
 Rules:
 - Assume good faith. Tab switches to documentation are normal engineering.
@@ -28,7 +31,7 @@ Rules:
 - Respond with ONLY a valid JSON object matching the provided schema —
   no markdown fences, no prose outside the JSON."""
 
-PROMPT_VERSION = "proctor-v2"
+PROMPT_VERSION = "proctor-v3"
 
 
 def render_context(learner_context) -> dict:

apps/ai-service/ai_service/prompts/variant.py

@@ -0,0 +1,46 @@
+"""Variant instantiation prompt (D-029, REQ-3-005).
+
+The model's ONLY job is to render already-sampled slot values into a task
+statement — it never invents parameters (the seeded sampler is pure code)
+and never changes difficulty. Prompt-injection surface is bounded: the
+variable inputs are the skeleton text, the seeded slot values, and the
+template title — nothing from the learner's environment.
+"""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+from ..llm.types import Message
+
+if TYPE_CHECKING:  # pragma: no cover - keeps this module pure text
+    from ..variants.templates import TaskTemplate
+
+VARIANT_SYSTEM_PROMPT = (
+    "You instantiate per-learner task variants for a competency-based AI school. "
+    "You receive a task statement skeleton and ALREADY-SAMPLED slot values. "
+    "Render the slot values into the skeleton, producing a complete, unambiguous "
+    "task statement a learner can build against. Rules:\n"
+    "- Use EXACTLY the given slot values; do not invent, rename, or add parameters.\n"
+    "- Keep the engineering depth IDENTICAL across draws: slot values change the "
+    "scenario, never the difficulty or scope.\n"
+    "- Keep the statement in the same language and register as the skeleton.\n"
+    "- Output STRICT JSON only: {\"statement\": \"<rendered statement>\"}.\n"
+)
+
+VARIANT_SCHEMA_HINT = '{"statement": "<complete rendered task statement string>"}'
+
+
+def render_variant_prompt(template: TaskTemplate, params: dict[str, str | int]) -> list[Message]:
+    """Messages for one seeded instantiation (D-020 defense drives the call)."""
+    slot_lines = "\n".join(f"  {{{slot.name}}} = {params[slot.name]!r}" for slot in template.slots)
+    user = (
+        f"Template: {template.title} (id={template.id})\n"
+        f"Statement skeleton:\n{template.statement_skeleton}\n\n"
+        f"Seeded slot values (use EXACTLY these):\n{slot_lines}\n\n"
+        "Render the complete task statement now."
+    )
+    return [
+        Message(role="system", content=VARIANT_SYSTEM_PROMPT),
+        Message(role="user", content=user),
+    ]

apps/ai-service/ai_service/sandbox/backend.py

@@ -46,7 +46,14 @@ class SandboxHandle(BaseModel):
 
 
 class SandboxSpec(BaseModel):
-    """Immutable description of the sandbox to lay out on disk."""
+    """Immutable description of the sandbox to lay out on disk.
+
+    `capture_env` (REQ-3-003): when non-empty, the backend starts a persistent
+    telemetry-wired sandbox — helper + inner namespaces + the stdlib capture
+    agent, launched with these env vars (NC_LEARNER_ID, NC_TASK_ID,
+    NC_INGEST_URL, NC_SANDBOX_ID). When None (default), spawn keeps the pure
+    shell semantics (REQ-3-001): disk layout only, fresh namespaces per exec.
+    """
 
     model_config = ConfigDict(arbitrary_types_allowed=True)
 
@@ -54,6 +61,7 @@ class SandboxSpec(BaseModel):
     learner_id: str
     workdir: Path
     limits: ResourceLimits = ResourceLimits()
+    capture_env: dict[str, str] | None = None
 
 
 class ExecResult(BaseModel):

apps/ai-service/ai_service/sandbox/manager.py

@@ -142,8 +142,18 @@ class SandboxManager:
 
     # -- lifecycle ----------------------------------------------------------
 
-    async def create(self, learner_id: str) -> SandboxHandleInfo:
-        """Spawn a sandbox for `learner_id`, or raise `PoolFullError` (D-032)."""
+    async def create(
+        self, learner_id: str, task_id: str | None = None
+    ) -> SandboxHandleInfo:
+        """Spawn a sandbox for `learner_id`, or raise `PoolFullError` (D-032).
+
+        `task_id` (REQ-3-003): when set, the sandbox is telemetry-wired — the
+        backend copies `scripts/sandbox-agent.py` into the workdir and starts
+        the stdlib capture agent inside the sandbox with the `NC_*` env baked
+        here (identity + WS ingest URL). The agent's lifecycle is tied to the
+        sandbox: `destroy()` reaps it (agent → inner → helper). When `task_id`
+        is None the sandbox is a pure shell sandbox (no capture).
+        """
         async with self._lock:
             if len(self._handles) >= self._settings.sandbox_max_concurrent:
                 raise PoolFullError(
@@ -153,13 +163,51 @@ class SandboxManager:
                 )
             sandbox_id = f"sbx-{uuid.uuid4().hex[:12]}"
             spec = workdir_mod.spec_for(sandbox_id, learner_id, self._settings)
+            if task_id is not None:
+                spec = spec.model_copy(
+                    update={
+                        "capture_env": self._capture_env(sandbox_id, learner_id, task_id)
+                    }
+                )
             handle = await self._backend.spawn(spec)
             self._handles[handle.id] = handle
             self._learner_ids[handle.id] = learner_id
             self._write_pid_marker(handle, learner_id)
-            logger.info("sandbox created: id=%s learner=%s", handle.id, learner_id)
+            logger.info(
+                "sandbox created: id=%s learner=%s task=%s",
+                handle.id,
+                learner_id,
+                task_id or "-",
+            )
             return self._info_for(handle)
 
+    def _capture_env(self, sandbox_id: str, learner_id: str, task_id: str) -> dict[str, str]:
+        """Env baked for the in-sandbox capture agent (REQ-3-003).
+
+        The agent joins the sandbox mount namespace but NOT its (offline)
+        network namespace, so it reaches this service over loopback
+        (`telemetry_ingest_host`, A-004 port).
+        """
+        from urllib.parse import urlencode
+
+        query = urlencode(
+            {
+                "learner_id": learner_id,
+                "task_id": task_id,
+                "sandbox_id": sandbox_id,
+            }
+        )
+        ingest_url = (
+            f"ws://{self._settings.telemetry_ingest_host}:{self._settings.port}"
+            f"/v1/telemetry/ingest?{query}"
+        )
+        return {
+            "NC_LEARNER_ID": learner_id,
+            "NC_TASK_ID": task_id,
+            "NC_SANDBOX_ID": sandbox_id,
+            "NC_INGEST_URL": ingest_url,
+        }
+
     async def list(self) -> list[SandboxHandleInfo]:
         """All live sandboxes (idle + busy; the backend has no busy flag)."""
         async with self._lock:

apps/ai-service/ai_service/sandbox/unshare_backend.py

@@ -1,24 +1,42 @@
 """UnshareBackend — D-024 Linux-namespace sandboxing via util-linux `unshare`.
 
-Each `exec` spawns:
+Two execution modes share one backend:
 
-    unshare --user --map-root-user --mount --pid --fork --net  sh -c '<shim>'
+1. Pure shell sandbox (`task_id is None`, REQ-3-001): isolation is established
+   PER-EXEC — every `exec` spawns a fresh namespace:
 
-The in-namespace shim (this util-linux build, 2.38, has no `unshare --bind`,
-so the bind happens as the first mount op inside the namespace) is:
+       unshare --user --map-root-user --mount --pid --fork --net  sh -c '<shim>'
 
-    mount -t tmpfs tmpfs /tmp            # private scratch, discarded on exit
-    mkdir -p /tmp/work
-    mount --bind <host workspace> /tmp/work
-    cd /tmp/work
-    ulimit -v/-t/-f  …                   # applied AFTER the bind, so rlimits
-    exec <cmd>                           # constrain the PAYLOAD, not unshare
+   There is no persistent process; the in-namespace shim is:
 
-Guarantees after this shim:
-  * uid 0 inside (mapped to the unprivileged host UID outside)
-  * no network: the fresh net namespace has no `lo` and no veth — zero links
-  * writes under `/work` land in the per-sandbox host workspace dir
-  * rlimits (RLIMIT_AS / RLIMIT_CPU / RLIMIT_FSIZE) constrain the payload only
+       mount -t tmpfs tmpfs /tmp            # private scratch, discarded on exit
+       mkdir -p /tmp/work
+       mount --bind <host workspace> /tmp/work
+       cd /tmp/work
+       ulimit -v/-t/-f  …                   # applied AFTER the bind, so rlimits
+       exec <cmd>                           # constrain the PAYLOAD, not unshare
+
+2. Telemetry-wired task sandbox (REQ-3-003, `capture_env` set): a PERSISTENT,
+   TRACKED topology so the stdlib capture agent can live inside the sandbox and
+   still stream events to ai-service. Per exec a fresh OFFLINE namespace would
+   leave the agent nowhere to run and (on this host, where a userns can't
+   bring `lo` up) no loopback to reach `ws://127.0.0.1`. So spawn creates a
+   long-lived helper (outer user+mount ns, ONLINE) and an inner sandbox
+   (mount+pid+fork+net — OFFLINE), both rooted at a private `ns/` subtree:
+
+       helper : unshare --user --map-root-user --mount  (mounts ns/ private)
+       inner  : unshare --mount --pid --fork --net      (tmpfs on ns/, bind
+                <workdir>/host/workspace -> <ns>/work)   <- the sandbox
+       exec   : nsenter -t <inner sleep> -m -- sh -c …  (joins inner mount ns;
+                offline + pid-isolated, uid 0, writes land on the host workspace)
+       agent  : nsenter -t <inner sleep> -m -- python3 <agent>  (joins the inner
+                MOUNT ns only — NOT pid/net — so it watches the live workspace
+                and stays ONLINE, reaching the app's WS ingest on loopback)
+
+   The agent is deliberately pid/net-exempt from the sandbox: it is OUR trusted
+   capture process, and isolating its network would cut the very link it needs.
+   `destroy` reaps agent → inner → helper (in that order). The handle's `pid`
+   is the agent's host pid (None for a pure shell sandbox).
 
 Why rlimits are applied in the shim, not Python's preexec_fn: setting
 RLIMIT_AS on the *unshare* process itself can trip the memory ceiling on the
@@ -30,16 +48,18 @@ Containment honesty (D-024 / G-1): a user namespace is NOT a write barrier.
 Writes made OUTSIDE the bind fall through to host paths, and because inner
 uid 0 maps to the invoking host uid, a sandboxed process can write anywhere
 that host uid can write. Isolation here is: private PIDs/MNT/NET/UTS, tmpfs
-scratch at /tmp, payload rlimits, and a uid map yielding no privilege the
-host uid did not already have. A per-sandbox runtime uid (D-025) is the
-follow-up that hardens DAC.
+scratch, payload rlimits, and a uid map yielding no privilege the host uid
+did not already have. A per-sandbox runtime uid (D-025) is the follow-up that
+hardens DAC.
 """
 
 from __future__ import annotations
 
 import asyncio
+import os
 import shlex
 import shutil
+import signal
 import time
 from datetime import UTC, datetime
 from pathlib import Path
@@ -48,22 +68,27 @@ from .backend import ExecResult, ResourceLimits, SandboxHandle, SandboxSpec
 from .workdir import create_layout
 from .workdir import snapshot as workdir_snapshot
 
-#: Args shared by every namespace we spawn (D-024). No `unshare --bind` on
-#: util-linux 2.38 — the bind is done from inside the namespace instead.
+#: Args shared by every PURE shell namespace we spawn (D-024). No
+#: `unshare --bind` on util-linux 2.38 — the bind is done from inside instead.
 UNSHARE_ARGS: tuple[str, ...] = (
     "--user",  # new user namespace …
     "--map-root-user",  # … in which we are uid 0 (mapped to host uid outside)
     "--mount",  # private mount table
     "--pid",  # private PID table
     "--fork",  # child is PID 1 in its namespace (reaps zombies, gets signals)
-    "--net",  # fresh net namespace: no lo, no veth → fully offline
+    "--net",  # fresh net namespace: no usable route → effectively offline
 )
 
-IN_NS_WORKDIR = "/tmp/work"  # where the workspace is bound inside the namespace
+IN_NS_WORKDIR = "/tmp/work"  # where the workspace is bound inside a pure shell ns
+
+#: Sentinels the long-lived namespace supervisors print once their mounts are
+#: laid out. exec()/the manager must not run before the bind exists.
+_HELPER_READY = "NC_HELPER_READY"
+_INNER_READY = "NC_INNER_READY"
 
 
 class SandboxUnavailableError(RuntimeError):
-    """`unshare` missing or user namespaces blocked on this host."""
+    """`unshare`/`nsenter` missing or user namespaces blocked on this host."""
 
 
 def _build_shim(workspace: Path, limits: ResourceLimits, cmd: list[str]) -> str:
@@ -90,35 +115,286 @@ def _build_shim(workspace: Path, limits: ResourceLimits, cmd: list[str]) -> str:
     )
 
 
+class _Tracked:
+    """The process tree + paths for one persistent (telemetry-wired) sandbox."""
+
+    def __init__(
+        self,
+        *,
+        helper: asyncio.subprocess.Process,
+        inner: asyncio.subprocess.Process,
+        agent: asyncio.subprocess.Process | None,
+        inner_pid: int,  # host pid of the SANDBOXED init (sleep) — ns enter target
+        host_dir: Path,
+        workspace: Path,
+        ns_root: Path,
+        ns_workdir: Path,
+    ) -> None:
+        self.helper = helper
+        self.inner = inner
+        self.agent = agent
+        self.inner_pid = inner_pid
+        self.host_dir = host_dir
+        self.workspace = workspace
+        self.ns_root = ns_root
+        self.ns_workdir = ns_workdir
+
+
 class UnshareBackend:  # satisfies SandboxBackend structurally (Protocol)
-    """D-024 backend: subprocess-per-exec inside fresh Linux namespaces."""
+    """D-024 backend: namespace subprocesses; persistent tree for task sandboxes."""
 
     def __init__(
         self,
         unshare_path: str | None = None,
         limits: ResourceLimits | None = None,  # per-spec override lands in 1-04
+        nsenter_path: str | None = None,
+        agent_script: Path | None = None,
     ) -> None:
         self._unshare = unshare_path or shutil.which("unshare") or "unshare"
+        self._nsenter = nsenter_path or shutil.which("nsenter") or "nsenter"
         self._limits = limits or ResourceLimits()
+        # The stdlib-only capture agent script, copied into each tracked
+        # workdir's host/ tree so nsenter can reach it inside the sandbox.
+        # ai_service/sandbox/unshare_backend.py -> parents[2] = apps/ai-service.
+        self._agent_script = agent_script or (
+            Path(__file__).resolve().parents[2] / "scripts" / "sandbox-agent.py"
+        )
+        # Tracked (persistent) sandboxes by id; pure shell sandboxes are absent.
+        self._tracked: dict[str, _Tracked] = {}
+
+    # -- spawn ------------------------------------------------------------------
 
     async def spawn(self, spec: SandboxSpec) -> SandboxHandle:
-        """Lay out the workdir and return a handle.
+        """Lay out the workdir; if `spec.capture_env` is set, start the sandbox.
 
-        Isolation is established per-`exec` (each exec = fresh namespaces), so
-        spawn only prepares on-disk state; there is no long-lived init process.
+        A spec WITHOUT capture_env keeps REQ-3-001 semantics: spawn only
+        prepares disk state and each exec forks a fresh (offline) namespace.
+        A spec WITH capture_env starts the persistent helper/inner tree and the
+        capture agent, and `handle.pid` carries the agent's host pid.
         """
         create_layout(spec)
+        if not spec.capture_env:
+            return SandboxHandle(
+                id=spec.sandbox_id,
+                pid=None,  # no persistent process; each exec forks short-lived PIDs
+                workdir=spec.workdir,
+                created_at=datetime.now(UTC),
+            )
+        tracked = await self._spawn_tracked(spec)
+        self._tracked[spec.sandbox_id] = tracked
         return SandboxHandle(
             id=spec.sandbox_id,
-            pid=None,  # no persistent process; each exec forks short-lived PIDs
+            pid=tracked.agent.pid if tracked.agent is not None else tracked.inner_pid,
             workdir=spec.workdir,
             created_at=datetime.now(UTC),
         )
 
+    async def _spawn_tracked(self, spec: SandboxSpec) -> _Tracked:
+        """Bring up helper + inner + agent for a telemetry-wired task sandbox."""
+        host_dir = spec.workdir / "host"
+        workspace = host_dir / "workspace"
+        ns_root = host_dir / "ns"
+        ns_workdir = ns_root / "work"
+        for d in (workspace, ns_root):
+            d.mkdir(parents=True, exist_ok=True)
+        # The agent script must live INSIDE the workspace: the inner ns bind
+        # mounts <host_dir>/workspace -> <ns_root>/work, so only workspace
+        # content is visible in-namespace at /work.
+        agent_host_path = workspace / "sandbox-agent.py"
+        shutil.copyfile(self._agent_script, agent_host_path)
+
+        helper = await self._launch_ns(
+            [
+                self._unshare,
+                "--user",
+                "--map-root-user",
+                "--mount",
+                "sh",
+                "-c",
+                (
+                    # Isolate ns/ so the inner tmpfs never propagates back to the
+                    # host mount table (make-private is best-effort on this host).
+                    f"mount --bind {shlex.quote(str(ns_root))} {shlex.quote(str(ns_root))}; "
+                    f"mount --make-private {shlex.quote(str(ns_root))} 2>/dev/null; "
+                    f"echo {_HELPER_READY}; exec sleep 3600"
+                ),
+            ],
+            sentinel=_HELPER_READY,
+            label="helper",
+        )
+        try:
+            inner = await self._launch_ns(
+                [
+                    *self._helper_join_argv(helper),
+                    self._unshare,
+                    "--mount",
+                    "--pid",
+                    "--fork",
+                    "--net",
+                    "sh",
+                    "-c",
+                    (
+                        f"mount -t tmpfs tmpfs {shlex.quote(str(ns_root))}; "
+                        f"mkdir -p {shlex.quote(str(ns_workdir))}; "
+                        f"mount --bind {shlex.quote(str(workspace))} "
+                        f"{shlex.quote(str(ns_workdir))}; "
+                        f"echo {_INNER_READY}; exec sleep 3600"
+                    ),
+                ],
+                sentinel=_INNER_READY,
+                label="inner",
+            )
+        except Exception:
+            await self._reap(helper)
+            raise
+
+        await asyncio.sleep(0)  # let the inner child's sleep fork settle
+        inner_pid = await asyncio.to_thread(self._find_child_pid, inner.pid)
+        if inner_pid is None:
+            await self._reap(inner)
+            await self._reap(helper)
+            raise SandboxUnavailableError(
+                f"could not resolve sandboxed init pid for {spec.sandbox_id}"
+            )
+
+        tracked = _Tracked(
+            helper=helper,
+            inner=inner,
+            agent=None,
+            inner_pid=inner_pid,
+            host_dir=host_dir,
+            workspace=workspace,
+            ns_root=ns_root,
+            ns_workdir=ns_workdir,
+        )
+        if spec.capture_env:
+            tracked.agent = await self._launch_agent(spec, tracked, agent_host_path)
+        return tracked
+
+    # -- process launch helpers --------------------------------------------------
+
+    def _helper_join_argv(self, helper: asyncio.subprocess.Process) -> list[str]:
+        """nsenter argv that runs a command inside the helper's user+mount ns."""
+        if helper.pid is None:
+            raise SandboxUnavailableError("helper namespace process is not running")
+        return [
+            self._nsenter,
+            "-t",
+            str(helper.pid),
+            "-m",
+            "-U",
+            "--preserve-credentials",
+            "--",
+        ]
+
+    def _sandbox_join_argv(self, tracked: _Tracked) -> list[str]:
+        """nsenter argv that joins the inner sandbox MOUNT namespace (uid 0)."""
+        return [self._nsenter, "-t", str(tracked.inner_pid), "-m", "--"]
+
+    async def _launch_agent(
+        self, spec: SandboxSpec, tracked: _Tracked, agent_host_path: Path
+    ) -> asyncio.subprocess.Process:
+        """Launch the capture agent: joins the sandbox mount ns, NOT pid/net.
+
+        The nsenter chain swaps the mount table under the process, so a HOST
+        cwd/relative path is invalid after the join (observed: python3
+        resolved ``sandbox-agent.py`` against a stale root → ``//…`` and
+        exited rc=2). The launch therefore happens through ``sh -c`` INSIDE
+        the joined namespace, using only in-namespace absolute paths: the
+        workspace is bind-mounted at ``<ns_root>/work``, the agent script was
+        copied into the host workspace, so ``/work/sandbox-agent.py`` exists
+        after the join. The agent runs ONLINE (joins mount ns only, not the
+        offline net ns) so it can dial the ai-service WS ingest loopback.
+        """
+        env = dict(spec.capture_env or {})
+        in_ns_script = f"{tracked.ns_root / 'work' / 'sandbox-agent.py'}"
+        in_ns_cwd = f"{tracked.ns_root / 'work'}"
+        launch = f"cd {shlex.quote(in_ns_cwd)} && exec python3 {shlex.quote(in_ns_script)}"
+        try:
+            return await asyncio.create_subprocess_exec(
+                *self._helper_join_argv(tracked.helper),
+                *self._sandbox_join_argv(tracked),
+                "sh",
+                "-c",
+                launch,
+                stdin=asyncio.subprocess.DEVNULL,
+                stdout=asyncio.subprocess.DEVNULL,
+                stderr=asyncio.subprocess.DEVNULL,
+                env=env,
+            )
+        except FileNotFoundError as exc:  # pragma: no cover - env-dependent
+            raise SandboxUnavailableError("python3 unavailable for capture agent") from exc
+
+    async def _launch_ns(
+        self, argv: list[str], *, sentinel: str, label: str
+    ) -> asyncio.subprocess.Process:
+        """Spawn a namespace supervisor and wait for its `sentinel` line."""
+        proc = await asyncio.create_subprocess_exec(
+            *argv,
+            stdout=asyncio.subprocess.PIPE,
+            stderr=asyncio.subprocess.STDOUT,
+        )
+
+        async def _wait_ready() -> None:
+            if proc.stdout is None:  # pragma: no cover (stdout is a PIPE)
+                raise SandboxUnavailableError(f"{label} namespace missing stdout pipe")
+            async for raw in proc.stdout:
+                if raw.decode(errors="replace").strip() == sentinel:
+                    return
+            raise SandboxUnavailableError(
+                f"{label} namespace exited before signalling readiness: {argv[:3]}"
+            )
+
+        try:
+            await asyncio.wait_for(_wait_ready(), timeout=10.0)
+        except TimeoutError as exc:
+            await self._reap(proc)
+            raise SandboxUnavailableError(
+                f"{label} namespace never became ready (timeout): {argv[:3]}"
+            ) from exc
+        except SandboxUnavailableError:
+            await self._reap(proc)
+            raise
+        return proc
+
+    @staticmethod
+    def _find_child_pid(parent_pid: int | None) -> int | None:
+        """First direct child of `parent_pid` (the pid-namespaced `sleep`).
+
+        The helper→unshare shim is inner.pid's parent chain head, but the
+        SANDBOXED mount/pid namespaces belong to its forked child (the
+        `sleep`). nsenter must target THAT pid to land inside the sandbox.
+        Reads /proc directly — best-effort, host-local, no subprocess.
+        """
+        if parent_pid is None:
+            return None
+        for entry in os.listdir("/proc"):
+            if not entry.isdigit():
+                continue
+            try:
+                with open(f"/proc/{entry}/stat") as fh:
+                    # ppid is field 4; comm (field 2) may contain spaces, so
+                    # parse relative to the LAST ')'.
+                    rest = fh.read().rsplit(") ", 1)[1].split()
+                if int(rest[1]) == parent_pid:  # state=rest[0], ppid=rest[1]
+                    return int(entry)
+            except (OSError, IndexError, ValueError):
+                continue
+        return None
+
+    # -- exec --------------------------------------------------------------------
+
     async def exec(self, handle: SandboxHandle, cmd: list[str]) -> ExecResult:
-        """Run `cmd` in a fresh namespace rooted at the sandbox workspace."""
+        """Run `cmd` in the sandbox workspace (cwd = the bound workspace)."""
         if not cmd:
             raise ValueError("exec requires a non-empty cmd")
+        tracked = self._tracked.get(handle.id)
+        if tracked is not None:
+            return await self._exec_tracked(handle, tracked, cmd)
+        return await self._exec_fresh(handle, cmd)
+
+    async def _exec_fresh(self, handle: SandboxHandle, cmd: list[str]) -> ExecResult:
+        """Pure shell sandbox: spawn one fresh offline namespace per exec."""
         workspace = handle.workdir / "workspace"
         if not workspace.is_dir():
             raise SandboxUnavailableError(f"spawn() first: no workspace at {workspace}")
@@ -129,7 +405,6 @@ class UnshareBackend:  # satisfies SandboxBackend structurally (Protocol)
             "-c",
             _build_shim(workspace, self._limits, cmd),
         ]
-
         started = time.monotonic()
         proc = await asyncio.create_subprocess_exec(
             *argv,
@@ -146,13 +421,128 @@ class UnshareBackend:  # satisfies SandboxBackend structurally (Protocol)
             duration_s=time.monotonic() - started,
         )
 
+    async def _exec_tracked(
+        self, handle: SandboxHandle, tracked: _Tracked, cmd: list[str]
+    ) -> ExecResult:
+        """Task sandbox: join the persistent inner namespace (offline, uid 0).
+
+        rlimits apply in the joining subshell so only the payload is limited;
+        cwd is the bound workspace (`<ns>/work`).
+        """
+        if tracked.inner.returncode is not None:
+            raise SandboxUnavailableError(
+                f"sandbox {handle.id} is not running (inner namespace exited)"
+            )
+        quoted_cmd = " ".join(shlex.quote(part) for part in cmd)
+        rlimit_prefix = (
+            f"ulimit -v {self._limits.memory_bytes // 1024}; "
+            f"ulimit -t {self._limits.cpu_seconds}; "
+            f"ulimit -f {self._limits.file_size_bytes // 512}; "
+        )
+        shell = (
+            f"cd {shlex.quote(str(tracked.ns_workdir))}; "
+            f"{rlimit_prefix}"
+            f"exec {quoted_cmd}"
+        )
+        argv = [
+            *self._helper_join_argv(tracked.helper),
+            *self._sandbox_join_argv(tracked),
+            "sh",
+            "-c",
+            shell,
+        ]
+        started = time.monotonic()
+        proc = await asyncio.create_subprocess_exec(
+            *argv,
+            stdin=asyncio.subprocess.DEVNULL,
+            stdout=asyncio.subprocess.PIPE,
+            stderr=asyncio.subprocess.PIPE,
+        )
+        out, err = await proc.communicate()
+        return ExecResult(
+            cmd=cmd,
+            returncode=proc.returncode if proc.returncode is not None else -1,
+            stdout=out.decode(errors="replace"),
+            stderr=err.decode(errors="replace"),
+            duration_s=time.monotonic() - started,
+        )
+
+    # -- snapshot / destroy --------------------------------------------------------
+
     async def snapshot(self, handle: SandboxHandle) -> Path:
-        return workdir_snapshot(handle.workdir)
+        workspace_root = handle.workdir
+        tracked = self._tracked.get(handle.id)
+        if tracked is not None:
+            # Copy the tracked workspace, not the legacy <workdir>/workspace.
+            dest_parent = handle.workdir / "snapshots"
+            dest_parent.mkdir(parents=True, exist_ok=True)
+            return workdir_snapshot_from_workspace(tracked.workspace, dest_parent)
+        return workdir_snapshot(workspace_root)
 
     async def destroy(self, handle: SandboxHandle) -> None:
-        """Best-effort teardown. Namespaces die with their process; nothing to kill.
+        """Best-effort teardown. Pure shell sandboxes die with their exec; for a
+        tracked task sandbox reap AGENT → INNER → HELPER so no capture process
+        or namespace supervisor outlives the handle (REQ-3-003 lifecycle).
 
         Keeping the workdir is deliberate: snapshots must survive destroy so a
         learner's last state can be restored by the manager layer.
         """
+        tracked = self._tracked.pop(handle.id, None)
+        if tracked is not None:
+            # Agent first (it must not flush a "stopped" event into a dead
+            # sandbox), then the namespace tree. The inner `unshare --fork`
+            # shim is NOT the namespace init: killing it orphans its child
+            # (the `sleep` that is PID 1 of the sandbox pid+mnt+net ns),
+            # which reparents to host init and holds the tmpfs + bind for
+            # a full hour (observed: ~30 leaked `sleep 3600` after a test
+            # run). `--kill-child` does not reach it either (util-linux
+            # 2.38 leaks the same child under this flag combo — the child
+            # is reparented before unshare's signal handler runs). The
+            # deterministic kill is SIGKILL on the ns-init's HOST pid,
+            # which we already track as `tracked.inner_pid` (nsenter uses
+            # it for exec); the kernel then tears down the namespace with
+            # its init (no processes remain).
+            for proc in (tracked.agent, tracked.inner, tracked.helper):
+                if proc is not None:
+                    await self._reap(proc)
+            self._kill_pid(tracked.inner_pid)
         handle.pid = None
+
+    @staticmethod
+    def _kill_pid(pid: int | None, sig: int = signal.SIGKILL) -> None:
+        """Best-effort host-side signal; pid recycled or gone is not an error."""
+        if pid is None:
+            return
+        try:
+            os.kill(pid, sig)
+        except (ProcessLookupError, PermissionError):
+            pass  # already dead, or not ours — nothing to do
+
+    @staticmethod
+    async def _reap(proc: asyncio.subprocess.Process) -> None:
+        """SIGTERM then SIGKILL, tolerant of an already-dead process."""
+        if proc.returncode is not None:
+            return
+        try:
+            proc.terminate()
+        except ProcessLookupError:
+            return
+        try:
+            await asyncio.wait_for(proc.wait(), timeout=5.0)
+        except TimeoutError:
+            try:
+                proc.kill()
+            except ProcessLookupError:
+                return
+            try:
+                await asyncio.wait_for(proc.wait(), timeout=5.0)
+            except TimeoutError:  # pragma: no cover - SIGKILL always wins
+                pass
+
+
+def workdir_snapshot_from_workspace(workspace: Path, snapshots_dir: Path) -> Path:
+    """Snapshot helper for tracked sandboxes whose workspace is `<workdir>/host/workspace`
+    instead of the legacy `<workdir>/workspace` layout."""
+    dest = snapshots_dir / datetime.now(UTC).strftime("%Y%m%dT%H%M%SZ")
+    shutil.copytree(workspace, dest, symlinks=False)
+    return dest

apps/ai-service/ai_service/telemetry/__init__.py

@@ -0,0 +1,21 @@
+"""Live build telemetry — event models and the TraceStore protocol (REQ-3-003).
+
+Boundary rule (D-027): telemetry/ imports from config only — never from
+agents/ or api/ (agents call engines through narrow interfaces, never
+the reverse; api/ composes stores via DI).
+"""
+
+from .ingest import IngestSession, TraceIntegrityMap, telemetry_ingest_endpoint
+from .models import EventKind, TelemetryEvent, TraceSpan
+from .store import SQLiteTraceStore, TraceStore
+
+__all__ = [
+    "EventKind",
+    "IngestSession",
+    "SQLiteTraceStore",
+    "TelemetryEvent",
+    "TraceIntegrityMap",
+    "TraceSpan",
+    "TraceStore",
+    "telemetry_ingest_endpoint",
+]

apps/ai-service/ai_service/telemetry/ingest.py

@@ -0,0 +1,448 @@
+"""WS ingest protocol for learner telemetry (REQ-3-003, D-026, G-3).
+
+Frame contract — trace identity travels as QUERY PARAMS on the WS upgrade
+(`WS /v1/telemetry/ingest?learner_id=...&task_id=...&sandbox_id=...`), NOT as
+a first init frame. Rationale: the in-sandbox capture agent (Task 2-2-01) is a
+stdlib-only RFC6455 client where the URL is the cheapest thing to parametrize
+(`NC_INGEST_URL` carries the query string); identity is also visible to the
+server BEFORE accept(), so a malformed handshake can be rejected without an
+accept/close round-trip. Client messages are then ONE event per JSON text
+frame — no envelope:
+
+    {"seq": 0, "kind": "command", "payload": {...}, "ts": "...",
+     "sandbox_id": "..."}   # learner_id / task_id forbidden (URL owns them)
+
+Server → client frames are typed status envelopes:
+
+    {"type": "ack_total",      "count": N}                 — final flush summary, then close 1000
+    {"type": "gap_warning",    "missing_seqs": [...]}      — seq skipped ahead
+    {"type": "event_rejected", "detail": "..."}       — one frame failed validation
+                                                        (seq echoed when parseable)
+    {"type": "event_rejected", "seq": N, "detail": "..."}  — stored-field rejected (bad kind)
+    {"type": "flooded",        "reason": "cap_exceeded"|"queue_overflow",
+     "count": N}                                           — sent before close(1008)
+
+Keepalive: the server sends an opaque ping frame every `PING_INTERVAL_S` (the
+capture agent auto-pongs at the frame layer); a peer that is silent past
+`PONG_TIMEOUT_S` is assumed wedged, but the keepalive half only LOGS — the
+receiver half owns disconnect detection (single-box pilot: TCP EOF is
+reliable; an aggressive pong-watchdog would false-positive on loaded boxes).
+
+Flood control (GRILL G-3, BINDING — silent drop-oldest is FORBIDDEN):
+  * per-connection inbound queue bounded at `INBOUND_QUEUE_MAX` frames; on
+    overflow → close code 1008 (policy violation) + trace marked
+    INCOMPLETE_FLOODED via `TraceIntegrityMap`.
+  * total events for the (learner, task) exceeding
+    `Settings.telemetry_max_events_per_task` → same 1008 + INCOMPLETE_FLOODED.
+  `INCOMPLETE_FLOODED` is an integrity signal Proctor/Phase-3 grader read via
+  `TraceIntegrityMap.is_incomplete()` (the G-4 gate): a flooded trace can
+  never yield a credential.
+
+Boundary (D-027): telemetry/ never imports agents/ or api/. This module
+imports only `fastapi.WebSocket` for the socket type (a protocol surface, not
+a DI framework); the session engine below depends only on the TraceStore
+protocol + Settings, and api/telemetry.py injects both through plain
+parameters.
+"""
+
+import asyncio
+import contextlib
+import json
+import logging
+from datetime import datetime
+from typing import Any, Final
+
+from fastapi import WebSocket, WebSocketDisconnect
+from pydantic import BaseModel, ConfigDict, Field, ValidationError
+
+from ..config import Settings
+from .models import TelemetryEvent
+from .store import TraceStore
+
+logger = logging.getLogger(__name__)
+
+#: WebSocket close code 1008 — policy violation (RFC 6455 §7.4.1).
+WS_CLOSE_POLICY_VIOLATION: Final = 1008
+
+#: Bounded inbound queue depth per connection (G-3). Sized for burst-tolerance
+#: well above the capture agent's emission rate; overflow is a flood signal,
+#: not a backpressure knob.
+INBOUND_QUEUE_MAX: Final = 256
+
+PING_INTERVAL_S: Final = 20.0
+
+
+class InboundEventFrame(BaseModel):
+    """Client → server event frame (one TelemetryEvent minus URL-owned ids).
+
+    `extra="forbid"`: learner_id/task_id arriving in the frame body is a
+    contract violation — identity comes from the query params only, so a
+    replayed frame can never lie about which trace it belongs to.
+    """
+
+    model_config = ConfigDict(extra="forbid")
+
+    seq: int = Field(ge=0)
+    kind: str = Field(min_length=1)
+    payload: dict[str, Any] = Field(default_factory=dict)
+    ts: datetime
+    sandbox_id: str = ""
+
+
+class TraceIntegrityMap:
+    """Integrity flags for traces that can never be graded (G-3/G-4).
+
+    Process-local and deliberately small: v0.3 runs ONE ai-service process per
+    box, and the Phase-3 grader reads this flag through the same DI container
+    — D-019-style in-memory registry precedent (the sandbox handle registry is
+    the same shape). The flag is terminal within the process: a reconnect
+    sending legal events does NOT clear it — the trace is already untrusted as
+    grading input. Restarting ai-service resets flags; grading runs against a
+    live service, and the SQLite trace rows themselves are durable.
+
+    All methods are sync: mutation is a dict write, reads are dict lookups —
+    no await needed, so callers from any layer (API handlers, the grader)
+    don't inherit an async surface for a nanosecond operation.
+    """
+
+    def __init__(self) -> None:
+        # (learner_id, task_id) -> machine-readable reason (INCOMPLETE_FLOODED)
+        self._flags: dict[tuple[str, str], str] = {}
+
+    def mark(self, learner_id: str, task_id: str, reason: str) -> None:
+        """Set an integrity flag. Presence of the flag is the signal; the
+        reason is informational (last write wins)."""
+        self._flags[(learner_id, task_id)] = reason
+
+    def clear(self, learner_id: str, task_id: str) -> None:
+        """Test seam: reset a flag (production ingest never clears)."""
+        self._flags.pop((learner_id, task_id), None)
+
+    def is_incomplete(self, learner_id: str, task_id: str) -> bool:
+        """True when the trace carries ANY terminal integrity flag."""
+        return (learner_id, task_id) in self._flags
+
+    def reason(self, learner_id: str, task_id: str) -> str | None:
+        """The flag's reason (INCOMPLETE_FLOODED), or None when unflagged."""
+        return self._flags.get((learner_id, task_id))
+
+
+class IngestSession:
+    """One WebSocket ingest connection: receive → queue → drain → store.
+
+    Two tasks per connection:
+      * `_receiver` — reads frames, validates shape, enqueues (bounded queue,
+        G-3). Receives never block on SQLite.
+      * `_drainer` — pops frames in arrival order, appends via TraceStore
+        (idempotent on (learner,task,seq)), emits gap warnings, enforces the
+        per-trace event cap.
+    Either task detecting a flood closes the WS with 1008 and marks the trace
+    INCOMPLETE_FLOODED. The events queue carries `None` as the client-
+    disconnect sentinel.
+
+    - `telemetry_max_events_per_task` is consulted at connect and re-checked
+      per append against the DURABLE row count (cap compares against stored
+      events, so a skipped-ahead seq cannot burn budget that was never sent).
+      Durable count via `TraceStore.count()` (COUNT(*)) — a single aggregate
+      per append, never materializing trace rows (the pre-P7 code read
+      `len(get_trace(...))` which was O(trace) per event / O(n²) per session).
+    """
+
+    def __init__(
+        self,
+        websocket: WebSocket,
+        store: TraceStore,
+        integrity: TraceIntegrityMap,
+        settings: Settings,
+        learner_id: str,
+        task_id: str,
+        sandbox_id: str,
+    ) -> None:
+        self._ws = websocket
+        self._store = store
+        self._integrity = integrity
+        # Snapshot of the one setting ingest consults: read once at connect so
+        # a hot-reloaded Settings object mid-session can't move the cap.
+        self._max_events = settings.telemetry_max_events_per_task
+        self.learner_id = learner_id
+        self.task_id = task_id
+        self.sandbox_id = sandbox_id
+
+        self._queue: asyncio.Queue[InboundEventFrame | None] = asyncio.Queue(
+            maxsize=INBOUND_QUEUE_MAX
+        )
+        self._seen: set[int] = set()
+        self._next_expected: int | None = None  # in-connection monotonic hint
+        self._received = 0
+        self._stored = 0
+        self._deduped = 0
+        self._rejected = 0
+        self._flooded = False
+        self._flood_reason = ""
+
+    # -- receive half ----------------------------------------------------------
+
+    async def run(self) -> None:
+        """Accept, run receiver+drainer, close cleanly. Owns the WS lifecycle."""
+        await self._ws.accept()
+        logger.info(
+            "telemetry ingest connected: %s/%s sandbox=%s",
+            self.learner_id,
+            self.task_id,
+            self.sandbox_id or "(none)",
+        )
+        pinger = asyncio.create_task(self._keepalive())
+        receiver = asyncio.create_task(self._receiver())
+        drainer = asyncio.create_task(self._drainer())
+        # First terminal outcome shuts the session down: client disconnect
+        # (receiver ends) → drainer flushes; drainer ended (clean close after
+        # flush or a 1008 flood close) → receiver must not linger.
+        pending: set[asyncio.Task[None]] = {receiver, drainer}
+        try:
+            done, pending = await asyncio.wait(
+                pending, return_when=asyncio.FIRST_COMPLETED
+            )
+            if receiver in done and drainer in pending:
+                try:
+                    await drainer  # final flush → sends ack_total, close 1000
+                finally:
+                    pending.discard(drainer)
+        finally:
+            for task in (pinger, *pending):
+                task.cancel()
+                with contextlib.suppress(asyncio.CancelledError):
+                    await task
+
+    async def _receiver(self) -> None:
+        """Read frames; parse+enqueue. Overflow → flood shutdown (G-3).
+
+        RuntimeError from receive_text is benign here: it fires when the
+        socket was closed by the drainer (1008 flood close) while this task
+        was parked in receive — a terminal condition, not a bug.
+        """
+        try:
+            while True:
+                raw = await self._ws.receive_text()
+                frame = self._parse(raw)
+                if frame is None:
+                    # Rejected frame — keep the connection open; the producer
+                    # gets an event_rejected status frame so a malformed batch
+                    # is visible (and its seq is never stored). Yield so the
+                    # status frame flushes before we block on the next receive.
+                    await self._reject_frame(raw)
+                    await asyncio.sleep(0)
+                    continue
+                self._received += 1
+                try:
+                    self._queue.put_nowait(frame)
+                except asyncio.QueueFull:
+                    # Bounded queue — overflow is a flood, never drop-oldest.
+                    # _trigger_flood closes the socket; fall through to the
+                    # tail so the disconnect sentinel is still enqueued — the
+                    # drainer is never left parked on an empty queue after a
+                    # flood (P7 review: the pre-fix code `return`ed from the
+                    # QueueFull branch WITHOUT the sentinel, leaking the
+                    # session task set — one per flooded trace).
+                    await self._trigger_flood("queue_overflow")
+                    return
+        except WebSocketDisconnect:
+            pass
+        except RuntimeError:
+            logger.debug(
+                "ingest receiver: socket already closed (flood path) %s/%s",
+                self.learner_id,
+                self.task_id,
+            )
+        # Client gone (clean close, drop, or flood close): sentinel unblocks
+        # the drainer for a final flush. put_nowait can only fail under flood,
+        # which already terminated the session.
+        with contextlib.suppress(asyncio.QueueFull):
+            self._queue.put_nowait(None)
+
+    def _parse(self, raw: str) -> InboundEventFrame | None:
+        """Validate one frame; None means malformed (caller rejects it)."""
+        try:
+            return InboundEventFrame.model_validate_json(raw)
+        except ValidationError:
+            return None
+
+    async def _reject_frame(self, raw: str) -> None:
+        """Malformed envelope: log + event_rejected status frame (never stored)."""
+        self._rejected += 1
+        detail = "invalid event frame"
+        try:
+            InboundEventFrame.model_validate_json(raw)
+        except ValidationError as exc:
+            detail = exc.errors()[0].get("msg", "validation error")
+        logger.warning(
+            "telemetry frame rejected: %s/%s: %s", self.learner_id, self.task_id, detail
+        )
+        seq: int | None = None
+        with contextlib.suppress(Exception):
+            seq = int(json.loads(raw).get("seq"))  # best-effort echo for the producer
+        payload: dict[str, Any] = {"type": "event_rejected", "detail": detail}
+        if seq is not None:
+            payload["seq"] = seq
+        await self._send_json(payload)
+
+    # -- drain half --------------------------------------------------------------
+
+    async def _drainer(self) -> None:
+        """Pop queued frames, append to the store, then close 1000 + summary."""
+        while True:
+            frame = await self._queue.get()
+            if frame is None:  # disconnect sentinel → flush complete
+                await self._send_json(
+                    {
+                        "type": "ack_total",
+                        "count": self._stored,
+                        "deduped": self._deduped,
+                        "rejected": self._rejected,
+                    }
+                )
+                with contextlib.suppress(RuntimeError, WebSocketDisconnect):
+                    await self._ws.close(code=1000)
+                return
+            await self._append(frame)
+
+    async def _append(self, frame: InboundEventFrame) -> None:
+        # Precedence: a trace already flagged INCOMPLETE_FLOODED is terminal —
+        # the connection that triggered it is being torn down, and any stray
+        # queued frames must not resurrect the trace's intake.
+        if self._integrity.is_incomplete(self.learner_id, self.task_id):
+            await self._trigger_flood("already_flagged")
+            return
+
+        # Per-trace cap (G-3): checked against the DURABLE row count so a
+        # reconnect resumes the budget instead of resetting it, and a
+        # skipped-ahead seq cannot burn budget that was never sent.
+        if self._flood_breached():
+            await self._trigger_flood("cap_exceeded")
+            return
+
+        # TelemetryEvent's @validates hooks fire on CONSTRUCTION (setattr), so
+        # the try must wrap building the model too — an unknown kind raises
+        # before `store.append` is ever reached.
+        event: TelemetryEvent
+        before = self._store.latest_seq(self.learner_id, self.task_id)
+        try:
+            event = TelemetryEvent(
+                learner_id=self.learner_id,
+                task_id=self.task_id,
+                seq=frame.seq,
+                kind=frame.kind,
+                payload=frame.payload,
+                ts=frame.ts,
+                sandbox_id=frame.sandbox_id or self.sandbox_id,
+            )
+            self._store.append(event)
+        except ValueError as exc:  # unknown kind / invalid field
+            self._rejected += 1
+            await self._send_json(
+                {"type": "event_rejected", "seq": frame.seq, "detail": str(exc)}
+            )
+            return
+        after = self._store.latest_seq(self.learner_id, self.task_id)
+
+        if after == before and frame.seq in self._seen:
+            self._deduped += 1  # at-least-once retry; stored once (idempotent)
+        else:
+            self._stored += 1
+        self._seen.add(frame.seq)
+        await self._check_gap(frame.seq)
+        # SQLite appends are sync and fast; on a burst the drainer can hold
+        # the loop between receives. Yield so the WS writer flushes the close
+        # and the pinger/interleave stay live under the eventlet-free portal.
+        await asyncio.sleep(0)
+
+    def _flood_breached(self) -> bool:
+        """True when this append would exceed the per-trace event budget."""
+        # Durable count (NOT latest_seq+1 — a skipped-ahead seq must not burn
+        # un-sent events' budget) via COUNT(*): never materialize the trace
+        # per append (P7 review — the old len(get_trace(...)) built every row
+        # object per event, O(trace) per append / O(n²) per session).
+        durable = self._store.count(learner_id=self.learner_id, task_id=self.task_id)
+        return durable >= self._max_events
+
+    async def _check_gap(self, incoming_seq: int) -> None:
+        """Seq skipped ahead → log + per-connection gap_warning status frame."""
+        if self._next_expected is not None and incoming_seq > self._next_expected:
+            missing = list(range(self._next_expected, incoming_seq))
+            logger.warning(
+                "telemetry gap: %s/%s missing seqs %s (arrived seq=%d)",
+                self.learner_id,
+                self.task_id,
+                missing,
+                incoming_seq,
+            )
+            await self._send_json({"type": "gap_warning", "missing_seqs": missing})
+        if self._next_expected is None or incoming_seq >= self._next_expected:
+            self._next_expected = incoming_seq + 1
+
+    # -- flood + keepalive ------------------------------------------------------
+
+    async def _trigger_flood(self, reason: str) -> None:
+        """G-3: 1008 close + INCOMPLETE_FLOODED mark. Exactly once."""
+        if self._flooded:
+            return
+        self._flooded = True
+        self._flood_reason = reason
+        self._integrity.mark(self.learner_id, self.task_id, "INCOMPLETE_FLOODED")
+        logger.warning(
+            "telemetry flood: %s/%s reason=%s — closing 1008, trace marked "
+            "INCOMPLETE_FLOODED (G-3; Proctor/grade gate will refuse it)",
+            self.learner_id,
+            self.task_id,
+            reason,
+        )
+        await self._send_json(
+            {"type": "flooded", "reason": reason, "count": self._received}
+        )
+        with contextlib.suppress(RuntimeError, WebSocketDisconnect):
+            await self._ws.close(
+                code=WS_CLOSE_POLICY_VIOLATION,
+                reason=f"telemetry flood control (G-3): {reason}",
+            )
+
+    async def _keepalive(self) -> None:
+        """Protocol-level ping on an interval (agent auto-pongs at frame level).
+
+        A send failure means the socket is already gone — the receiver half
+        independently surfaces the disconnect; we just stop pinging.
+        """
+        while True:
+            await asyncio.sleep(PING_INTERVAL_S)
+            try:
+                await self._ws.send_bytes(b"\x89ping-nextcraft")
+            except (RuntimeError, WebSocketDisconnect):
+                return
+
+    async def _send_json(self, payload: dict[str, Any]) -> None:
+        """Best-effort status frame; the socket may already be gone."""
+        with contextlib.suppress(RuntimeError, WebSocketDisconnect):
+            await self._ws.send_json(payload)
+
+
+async def telemetry_ingest_endpoint(
+    websocket: WebSocket,
+    learner_id: str,
+    task_id: str,
+    store: TraceStore,
+    integrity: TraceIntegrityMap,
+    settings: Settings,
+    sandbox_id: str = "",
+) -> None:
+    """Engine entry: build the session and run it. api/telemetry.py calls this
+    with query params + app.state services already resolved — this signature
+    is deliberately Depends-free (telemetry/ never knows FastAPI DI exists).
+    """
+    session = IngestSession(
+        websocket=websocket,
+        store=store,
+        integrity=integrity,
+        settings=settings,
+        learner_id=learner_id,
+        task_id=task_id,
+        sandbox_id=sandbox_id,
+    )
+    await session.run()

apps/ai-service/ai_service/telemetry/models.py

@@ -0,0 +1,108 @@
+"""Telemetry event record — the row the trace store persists (REQ-3-003, D-027).
+
+One model serves both as the JSON payload sent by producers and as the SQLite
+row schema. `payload` is stored as a JSON column (native JSONB on Postgres —
+no migration-time shape change, D-027).
+
+Field contract (consumed by the trace store and the grader):
+    learner_id  — non-empty learner identifier.
+    task_id     — non-empty task/session identifier; trace identity is the
+                  (learner_id, task_id) pair.
+    seq         — sequence number per trace, >= 0. Monotonicity per
+                  (learner, task) is enforced by the store (Task 2-1-02);
+                  this model only rejects negative seqs.
+    kind        — event discriminator: command | file_diff | run_result |
+                  test_result | activity | stdin | stdout.
+    payload     — free-form JSON detail blob.
+    ts          — envelope timestamp (UTC); monotonicity enforced at ingest.
+    sandbox_id  — originating sandbox ("" for non-sandbox sources).
+
+Boundary (D-027): telemetry/ never imports agents/ or api/.
+"""
+
+from datetime import datetime
+from typing import Any, Literal
+
+from pydantic import BaseModel, ConfigDict, Field
+from sqlalchemy import JSON, Index, String
+from sqlalchemy.orm import validates
+from sqlmodel import Field as SQLField
+from sqlmodel import SQLModel
+
+EventKind = Literal[
+    "command",
+    "file_diff",
+    "run_result",
+    "test_result",
+    "activity",
+    "stdin",
+    "stdout",
+]
+_EVENT_KINDS: frozenset[str] = frozenset(EventKind.__args__)
+
+
+class TelemetryEvent(SQLModel, table=True):
+    """A single durable telemetry event; (learner_id, task_id, seq) is PK.
+
+    Constraint enforcement uses SQLAlchemy `@validates` hooks: sqlmodel
+    0.0.42's metaclass drops pydantic `Field(ge=...)`/`field_validator`
+    constraints for table models (the decorators register but never make it
+    into the core schema), while `@validates` fires on every attribute set —
+    construction included — and raises ValueError on violation. seq >= 0 plus
+    a VARCHAR kind column keep the DB shape Postgres-ready (D-027).
+    """
+
+    __tablename__ = "telemetry_event"
+    # PK columns already produce a unique index; this secondary index covers
+    # trace reads ordered by seq without depending on the PK column order
+    # (Postgres migration target D-027).
+    __table_args__ = (Index("ix_telemetry_event_trace", "learner_id", "task_id"),)
+
+    learner_id: str = SQLField(primary_key=True)
+    task_id: str = SQLField(primary_key=True)
+    seq: int = SQLField(primary_key=True)
+    # Bare Literal annotations crash sqlmodel<=0.0.42's column inference
+    # (issubclass(TypeAlias, Enum)); an explicit sa_type + the validates hook
+    # below gives the same contract: VARCHAR column, Literal-rejected values.
+    kind: EventKind = SQLField(sa_type=String)
+    # JSON column: stored as TEXT on SQLite, native JSONB on Postgres (D-027).
+    payload: dict[str, Any] = SQLField(default_factory=dict, sa_type=JSON)
+    ts: datetime
+    sandbox_id: str = SQLField(default="")
+
+    @validates("learner_id", "task_id")
+    def _ids_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty identifier")
+        return value
+
+    @validates("seq")
+    def _seq_non_negative(self, key: str, value: int) -> int:
+        if value < 0:
+            raise ValueError("seq must be >= 0 (monotonicity is the store's job)")
+        return value
+
+    @validates("kind")
+    def _kind_is_known(self, key: str, value: str) -> str:
+        if value not in _EVENT_KINDS:
+            raise ValueError(f"unknown event kind: {value!r}")
+        return value
+
+
+class TraceSpan(BaseModel):
+    """Derived view: the ordered event trace for one (learner_id, task_id).
+
+    NOT a table — materialized by the store from persisted TelemetryEvents
+    (grader/Lab consume this shape; replay order is the seq column).
+    """
+
+    model_config = ConfigDict(frozen=True)
+
+    learner_id: str = Field(min_length=1)
+    task_id: str = Field(min_length=1)
+    events: tuple[TelemetryEvent, ...] = ()
+
+    @property
+    def latest_seq(self) -> int:
+        """Highest seq in the span; -1 when empty (store convention)."""
+        return self.events[-1].seq if self.events else -1

apps/ai-service/ai_service/telemetry/store.py

@@ -0,0 +1,218 @@
+"""TraceStore — telemetry persistence protocol + SQLite implementation (REQ-3-003, D-027).
+
+Postgres-migration-ready (D-027): the protocol is the only surface the API /
+grader layers touch; swapping SQLiteTraceStore for a Postgres-backed
+implementation must not change call sites. The `telemetry_event` table uses
+only portable column types (str / int / datetime / JSON), so the same SQLModel
+schema stands up unchanged on Postgres.
+
+Ingest is at-least-once: duplicates carry the same (learner_id, task_id, seq)
+idempotency key, so `append` with a triplet that is already stored is a no-op.
+The pair (learner_id, task_id) identifies a trace; `seq` numbers events in it
+starting at 0.
+
+Concurrency (a-3): the engine enables WAL + synchronous=NORMAL and a busy
+timeout at connection time, so the ingest writer and grader readers do not hit
+`database is locked` on the single-box pilot.
+
+Boundary: `telemetry/` never imports `agents/` / `api/` and has no FastAPI
+dependency.
+"""
+
+import logging
+import sqlite3
+from collections.abc import Iterator
+from contextlib import contextmanager
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any, Protocol
+
+import sqlalchemy as sa
+from sqlmodel import Session, SQLModel, create_engine, select
+
+from ..config import Settings
+from .models import TelemetryEvent
+
+logger = logging.getLogger(__name__)
+
+
+class TraceStore(Protocol):
+    """Persistence contract for ordered per-learner task trace streams.
+
+    Implemented by SQLiteTraceStore (v0.3, D-027); a Postgres implementation
+    must satisfy the same surface.
+    """
+
+    def append(self, event: TelemetryEvent) -> None:
+        """Store one event. IDEMPOTENT on (learner_id, task_id, seq):
+
+        at-least-once ingest retries with the same triplet are deduped
+        (stored once), not rejected. Later events must not overwrite an
+        existing row.
+        """
+        ...
+
+    def get_trace(self, learner_id: str, task_id: str) -> list[TelemetryEvent]:
+        """All stored events for the trace, ordered by seq ascending.
+
+        Detached from any DB session — safe to pass across layers. Empty list
+        when the trace has no events.
+        """
+        ...
+
+    def gaps(self, learner_id: str, task_id: str) -> list[int]:
+        """Missing seqs in 0..latest for the trace ([0,2,3] stored -> [1])."""
+        ...
+
+    def latest_seq(self, learner_id: str, task_id: str) -> int:
+        """Highest stored seq for the trace; -1 when no events exist."""
+        ...
+
+    def count(self, learner_id: str, task_id: str) -> int:
+        """Number of stored events for the trace (COUNT(*), never
+        materializes rows — the ingest cap consults this per append, so
+        an O(trace) implementation would make ingest O(n²) per session).
+        """
+        ...
+
+    def list_tasks(self, learner_id: str) -> list[str]:
+        """Distinct task_ids with at least one event for the learner."""
+        ...
+
+    def close(self) -> None:
+        """Release DB connections. Store must not be used after close."""
+        ...
+
+
+def _sqlite_connect(dbapi_connection: sqlite3.Connection, _: object) -> None:
+    """Per-connection pragma setup (a-3).
+
+    journal_mode=WAL  — readers never block the single writer.
+    synchronous=NORMAL — safe in WAL mode, avoids full fsync-per-commit.
+    busy_timeout=5000  — retry briefly under contention instead of
+                         `OperationalError: database is locked`.
+    """
+    cursor = dbapi_connection.cursor()
+    cursor.execute("PRAGMA journal_mode=WAL")
+    cursor.execute("PRAGMA synchronous=NORMAL")
+    cursor.execute("PRAGMA busy_timeout=5000")
+    cursor.close()
+
+
+def _as_utc(ts: datetime) -> datetime:
+    """Timestamps travel as naive datetime on SQLite, tz-aware elsewhere.
+
+    SQLite (via SQLModel) drops tzinfo; Postgres TIMESTAMP WITH TIME ZONE
+    keeps it. Normalizing on the read/write boundary makes the store's
+    contract tz-aware UTC regardless of the backend (D-027).
+    """
+    if ts.tzinfo is None:
+        return ts.replace(tzinfo=UTC)  # naive UTC read-side: label as UTC
+    return ts.astimezone(UTC)
+
+
+class SQLiteTraceStore:
+    """SQLite-backed TraceStore (SQLModel). First real persistence (D-027)."""
+
+    def __init__(self, db_path: Path | None = None) -> None:
+        self._db_path: Path = db_path if db_path is not None else Settings().db_path
+        self._engine = create_engine(f"sqlite:///{self._db_path}")
+        sa.event.listen(self._engine, "connect", _sqlite_connect)
+        SQLModel.metadata.create_all(self._engine)
+
+    @contextmanager
+    def _session(self) -> Iterator[Session]:
+        # expire_on_commit=False: ORM objects returned from `append`'s
+        # IntegrityError path stay usable without a refresh round-trip.
+        with Session(self._engine, expire_on_commit=False) as session:
+            yield session
+
+    def append(self, event: TelemetryEvent) -> None:
+        # INSERT-if-absent via PK: sqlite3 raises IntegrityError on a
+        # duplicate (learner_id, task_id, seq); swallow it — the row is
+        # already stored, which is the dedup contract for at-least-once
+        # ingest. `session.merge` would upsert instead; wrong semantics here.
+        with self._session() as session:
+            try:
+                session.add(event)
+                session.commit()
+            except sa.exc.IntegrityError:
+                session.rollback()
+                logger.debug(
+                    "trace event dedup: %s/%s seq=%d already stored",
+                    event.learner_id,
+                    event.task_id,
+                    event.seq,
+                )
+
+    def get_trace(self, learner_id: str, task_id: str) -> list[TelemetryEvent]:
+        with self._session() as session:
+            stmt = (
+                select(TelemetryEvent)
+                .where(TelemetryEvent.learner_id == learner_id)
+                .where(TelemetryEvent.task_id == task_id)
+                .order_by(TelemetryEvent.seq)
+            )
+            results = session.exec(stmt).all()
+            # Detach from the session: callers must not depend on open-session
+            # ORM magic (lazy loads fail once the session is closed).
+            for row in results:
+                row.ts = _as_utc(row.ts)
+                session.expunge(row)
+            return list(results)
+
+    def _stored_seqs(self, learner_id: str, task_id: str) -> list[int]:
+        with self._session() as session:
+            stmt = (
+                select(TelemetryEvent.seq)
+                .where(TelemetryEvent.learner_id == learner_id)
+                .where(TelemetryEvent.task_id == task_id)
+                .order_by(TelemetryEvent.seq)
+            )
+            # sqlmodel scalar select: rows are plain ints, not 1-tuples.
+            return [int(seq) for seq in session.exec(stmt).all()]
+
+    def gaps(self, learner_id: str, task_id: str) -> list[int]:
+        seqs = self._stored_seqs(learner_id, task_id)
+        if not seqs:
+            return []
+        present = set(seqs)
+        # seq numbering starts at 0; a gap is any seq in 0..latest not stored.
+        return [seq for seq in range(seqs[-1] + 1) if seq not in present]
+
+    def latest_seq(self, learner_id: str, task_id: str) -> int:
+        with self._session() as session:
+            stmt = (
+                select(sa.func.max(TelemetryEvent.seq))
+                .where(TelemetryEvent.learner_id == learner_id)
+                .where(TelemetryEvent.task_id == task_id)
+            )
+            latest: Any = session.exec(stmt).one()
+        return -1 if latest is None else int(latest)
+
+    def count(self, learner_id: str, task_id: str) -> int:
+        # COUNT(*) at the DB — no row materialization. The ingest flood cap
+        # calls this per append (telemetry/ingest._flood_breached); the
+        # docstring-free body keeps it obvious what the query shape is.
+        with self._session() as session:
+            stmt = (
+                select(sa.func.count(TelemetryEvent.seq))
+                .where(TelemetryEvent.learner_id == learner_id)
+                .where(TelemetryEvent.task_id == task_id)
+            )
+            total: Any = session.exec(stmt).one()
+        return int(total or 0)
+
+    def list_tasks(self, learner_id: str) -> list[str]:
+        with self._session() as session:
+            stmt = (
+                select(TelemetryEvent.task_id)
+                .where(TelemetryEvent.learner_id == learner_id)
+                .distinct()
+                .order_by(TelemetryEvent.task_id)
+            )
+            # sqlmodel scalar select: rows are plain strs, not 1-tuples.
+            return [str(task_id) for task_id in session.exec(stmt).all()]
+
+    def close(self) -> None:
+        self._engine.dispose()

apps/ai-service/ai_service/variants/__init__.py

@@ -0,0 +1,31 @@
+"""Per-learner variant task generation — templates, generator, VariantStore (REQ-3-005).
+
+Boundary rule (D-027): variants/ is an engine module — it never imports
+api/; its ONLY agents/ dependency is the module-direct
+agents.structured import in generator.py (the sanctioned shared D-020
+structured defense, same exception as grading/engine.py). api/ composes
+the generator and store via DI; store.py imports config only.
+
+CO-ORDINATION NOTE (ADD, don't REMOVE — same convention as grading/):
+This __init__.py is a minimal placeholder created by the VariantStore
+task (4-1-02). The templates task (4-1-01) owns this file's final shape
+— when templates.py lands, ADD its exports alongside these; do not
+remove the store exports below.
+
+Wave status: store.py (VariantRecord, VariantStore, SQLiteVariantStore)
+landed in Wave 1 (task 4-1-02); templates.py is Wave 1 task 4-1-01;
+generator.py is Wave 2 (4-2-01).
+"""
+
+from .store import SQLiteVariantStore, VariantRecord, VariantStore
+from .templates import TEMPLATES, TaskTemplate, get_template, template_for_competency
+
+__all__ = [
+    "SQLiteVariantStore",
+    "TEMPLATES",
+    "TaskTemplate",
+    "VariantRecord",
+    "VariantStore",
+    "get_template",
+    "template_for_competency",
+]

apps/ai-service/ai_service/variants/generator.py

@@ -0,0 +1,136 @@
+"""Seeded per-learner variant generator (D-029, REQ-3-005).
+
+Contract (binding, from GRILL + PLAN Must-Haves):
+- REPRODUCIBLE: seed = sha256(template_id|learner_id|milestone); the same
+  (template, learner) re-derives the same seed, params, task_id — and the
+  second generate() call is a cache hit with NO LLM call.
+- DISTINCT: different learners on the same template draw different params
+  (the sampler is seeded per-learner) and receive distinct statements.
+- NEVER BLOCKS ON THE LLM: the deterministic skeleton render
+  (`template.render(params)`) is a complete, valid statement; if the D-020
+  LLM render fails after its bounded retry, the fallback is used — and
+  because the fallback is exactly `template.render(seed-params)`, it is
+  auditable from the persisted seed + params without a provenance column.
+- AUDITABLE: seed + params + statement persist via VariantStore
+  (insert-only first-wins) — the proctoring cross-check path.
+- FAIR (a-5): slot draws change the scenario, never the difficulty; the
+  template's rubric anchors bound the expected effort envelope, so every
+  variant of one template is held to the same bar.
+"""
+
+from __future__ import annotations
+
+import hashlib
+from datetime import UTC, datetime
+from typing import TYPE_CHECKING
+
+from pydantic import BaseModel, ConfigDict, Field
+
+from ..agents.structured import StructuredOutputError, structured_completion
+from ..llm.types import Message
+from ..prompts.variant import VARIANT_SCHEMA_HINT, render_variant_prompt
+from .store import VariantRecord
+from .templates import TaskTemplate, get_template
+
+if TYPE_CHECKING:  # pragma: no cover
+    from ..llm.base import LLMProvider
+    from .store import VariantStore
+
+MILESTONE = "v0.3"
+
+
+class RenderedVariant(BaseModel):
+    """D-20-validated LLM render output (statement only — files come from the template)."""
+
+    model_config = ConfigDict(extra="forbid")
+
+    statement: str = Field(min_length=20)
+
+
+class UnknownTemplateError(ValueError):
+    """Raised when generate() is asked for a template id not in the library."""
+
+
+def derive_seed(template_id: str, learner_id: str, milestone: str = MILESTONE) -> str:
+    """Reproducible per-(template, learner, milestone) seed (D-029)."""
+    return hashlib.sha256(f"{template_id}|{learner_id}|{milestone}".encode()).hexdigest()
+
+
+def derive_task_id(seed: str) -> str:
+    """Deterministic grading/telemetry task key from the seed (16 hex chars)."""
+    return f"task-{seed[:16]}"
+
+
+class VariantGenerator:
+    """Seeded instantiation over the template library. DI: store + provider."""
+
+    def __init__(self, store: VariantStore, provider: LLMProvider, model: str) -> None:
+        self._store = store
+        self._provider = provider
+        self._model = model
+
+    async def generate(self, learner_id: str, template_id: str) -> VariantRecord:
+        template = get_template(template_id)
+        if template is None:
+            raise UnknownTemplateError(f"no task template with id {template_id!r}")
+
+        # Cache: D-029 reproducibility — same (learner, template) is served
+        # from the store with no LLM call.
+        cached = self._store.get(learner_id, template_id)
+        if cached is not None:
+            return cached
+
+        seed_hex = derive_seed(template_id, learner_id)
+        task_id = derive_task_id(seed_hex)
+        params = template.sample_params(_seed_int(seed_hex))
+        _validate_params(template, params)
+
+        statement = await self._render(template, params)
+        record = VariantRecord(
+            learner_id=learner_id,
+            task_id=task_id,
+            template_id=template_id,
+            seed=seed_hex,
+            params=dict(params),
+            statement=statement,
+            starter_files=dict(template.starter_files),
+            created_at=datetime.now(UTC),
+        )
+        self._store.save(record)
+        return record
+
+    async def _render(self, template: TaskTemplate, params: dict[str, str | int]) -> str:
+        """LLM render via D-020; deterministic fallback never blocks task work.
+
+        Provenance note: unlike grades, variants carry no `model` column —
+        the deterministic fallback is exactly `template.render(params)`,
+        re-derivable from the persisted seed + params, so a fallback render is
+        auditable without storing provenance (the seed IS the provenance).
+        """
+        messages: list[Message] = render_variant_prompt(template, params)
+        try:
+            rendered = await structured_completion(
+                self._provider,
+                messages,
+                model=self._model,
+                schema=RenderedVariant,
+                schema_hint=VARIANT_SCHEMA_HINT,
+            )
+        except StructuredOutputError:
+            # Deterministic fallback: the skeleton + seeded slots is already a
+            # complete statement, re-derivable from the persisted seed.
+            return template.render(params)
+        return rendered.statement
+
+
+def _seed_int(seed_hex: str) -> int:
+    """Stable int for random.Random from the hex seed."""
+    return int(seed_hex[:16], 16)
+
+
+def _validate_params(template: TaskTemplate, params: dict[str, str | int]) -> None:
+    """Defense in depth: every sampled value must be schema-valid (a-5)."""
+    for slot in template.slots:
+        value = params.get(slot.name)
+        if value is None or not slot.validate_value(value):
+            raise ValueError(f"sampled params invalid for slot {slot.name!r}: {value!r}")

apps/ai-service/ai_service/variants/store.py

@@ -0,0 +1,311 @@
+"""VariantStore — variant persistence protocol + SQLite implementation (REQ-3-005, D-027).
+
+Postgres-migration-ready (D-027): the protocol is the only surface the
+variant generator and API layers touch; swapping SQLiteVariantStore for a
+Postgres-backed implementation must not change call sites. The
+`variant_record` table uses only portable column types (str / JSON /
+datetime), so the same SQLModel schema stands up unchanged on Postgres.
+
+Insert-only, NOT upsert: (learner_id, template_id) is the variant identity
+and the FIRST generation is authoritative — reproducibility (D-029) means
+the seed re-derives the same variant, so the generator's cache path serves
+`get` instead of saving again. `save` is a plain INSERT; a duplicate pair
+raises sqlalchemy.exc.IntegrityError to the caller (documented behavior).
+`task_id` is unique too — it is the grading/telemetry trace key, so a
+trace or grade can never silently join to a different variant. Both
+rejections are deliberate: overwriting a stored variant would swap a
+learner's graded task underneath its trace and grade (audit corruption).
+Contrast TraceStore.append (dedup-keep-first, swallowed — at-least-once
+ingest) and GradeStore.save (upsert-latest-wins — a regrade is
+latest-state); this store is the third contract of the D-027 family.
+
+Concurrency (a-3): the store enables WAL + synchronous=NORMAL and a busy
+timeout at connection time, so a generation writer and API readers do not
+hit `database is locked` on the single-box pilot.
+
+`created_at` contract: callers stamp UTC (datetime.now(UTC)); SQLite
+stores it naive and the read paths re-label it tz-aware UTC (same
+boundary normalization as TelemetryEvent.ts / GradeRecord.created_at, so
+the contract holds on any backend).
+
+Boundary (D-027): `variants/` never imports `agents/` / `api/`; this
+module imports config only.
+"""
+
+import logging
+import sqlite3
+from collections.abc import Iterator
+from contextlib import contextmanager
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any, Protocol
+
+import sqlalchemy as sa
+from sqlalchemy import JSON, Index, UniqueConstraint
+from sqlalchemy.orm import validates
+from sqlmodel import Field, Session, SQLModel, create_engine, select
+
+from ..config import Settings
+
+logger = logging.getLogger(__name__)
+
+
+class VariantRecord(SQLModel, table=True):
+    """A persisted task variant; (learner_id, template_id) is the PK — first wins.
+
+    Written once by the variant generator (Task 4-2-01), read by the API
+    layer and proctoring cross-checks through the VariantStore protocol.
+    Constraint enforcement mirrors TelemetryEvent / GradeRecord: sqlmodel
+    0.0.42's metaclass drops pydantic constraints on table models, so
+    SQLAlchemy `@validates` hooks enforce instead and the column types
+    stay Postgres-ready (D-027).
+
+    Field contract:
+        learner_id     — non-empty learner identifier (same id space as
+                         traces and grades).
+        template_id    — non-empty task template identifier; variant
+                         identity is the (learner_id, template_id) pair —
+                         the pair the generator caches on (exactly one
+                         variant per learner per template).
+        task_id        — non-empty, GLOBALLY unique task identifier; the
+                         grading/telemetry trace key (the (learner_id,
+                         task_id) pair TraceStore / GradeStore key on),
+                         stamped at generation so a variant's trace and
+                         grade join back to it exactly once.
+        seed           — non-empty variant seed (D-029); derived from
+                         (template_id, learner_id, milestone) so the
+                         variant is reproducible and auditable.
+        params         — typed parameter-slot values the generator filled;
+                         JSON dict. An empty dict is legal (a slotless
+                         template).
+        statement      — non-empty rendered task statement shown to the
+                         learner (distinct per learner by construction,
+                         REQ-3-005).
+        starter_files  — workspace scaffold: filename -> file content;
+                         JSON dict. An empty dict is legal (no scaffold).
+        created_at     — UTC generation timestamp.
+    """
+
+    __tablename__ = "variant_record"
+    # The composite PK covers (learner_id, template_id) point lookups; the
+    # unique task_id covers get_by_task (the grading/telemetry join path);
+    # the two secondary indexes cover list_for_learner / list_by_template
+    # ordered by created_at without a sort step (Postgres target D-027).
+    __table_args__ = (
+        UniqueConstraint("task_id", name="uq_variant_record_task_id"),
+        Index("ix_variant_record_learner_created", "learner_id", "created_at"),
+        Index("ix_variant_record_template_created", "template_id", "created_at"),
+    )
+
+    learner_id: str = Field(primary_key=True)
+    template_id: str = Field(primary_key=True)
+    task_id: str
+    seed: str
+    # JSON columns: stored as TEXT on SQLite, native JSONB on Postgres (D-027).
+    params: dict[str, Any] = Field(default_factory=dict, sa_type=JSON)
+    statement: str
+    starter_files: dict[str, Any] = Field(default_factory=dict, sa_type=JSON)
+    created_at: datetime
+
+    @validates("learner_id", "template_id", "task_id")
+    def _ids_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty identifier")
+        return value
+
+    @validates("seed")
+    def _seed_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty seed string")
+        return value
+
+    @validates("statement")
+    def _statement_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty statement string")
+        return value
+
+
+class VariantStore(Protocol):
+    """Persistence contract for reproducible per-learner task variants.
+
+    Implemented by SQLiteVariantStore (v0.3, D-027); a Postgres
+    implementation must satisfy the same surface.
+    """
+
+    def save(self, variant: VariantRecord) -> None:
+        """Persist a new variant. INSERT-ONLY on (learner_id, template_id):
+        the FIRST generated variant is authoritative (reproducibility,
+        D-029); a duplicate pair raises sqlalchemy.exc.IntegrityError to
+        the caller — the generator serves cached variants via `get`
+        instead of saving again. `task_id` is unique too: claiming an
+        existing trace key for a different variant is equally rejected.
+        NOT upsert; contrast GradeStore.save (latest-wins) and
+        TraceStore.append (dedup-keep-first, swallowed).
+        """
+        ...
+
+    def get(self, learner_id: str, template_id: str) -> VariantRecord | None:
+        """The learner's stored variant for the template; None when none
+        exists. Detached from any DB session — safe to pass across layers.
+        """
+        ...
+
+    def get_by_task(self, task_id: str) -> VariantRecord | None:
+        """The variant owning the task key (the grading/telemetry join
+        path); None when none exists. Detached from any DB session.
+        """
+        ...
+
+    def list_for_learner(self, learner_id: str) -> list[VariantRecord]:
+        """All stored variants for the learner, ordered by created_at
+        ascending (chronological; task_id breaks same-instant ties).
+        Empty list when the learner has none.
+        """
+        ...
+
+    def list_by_template(self, template_id: str) -> list[VariantRecord]:
+        """All stored variants generated from the template — one row per
+        learner — ordered by created_at ascending (chronological;
+        learner_id breaks same-instant ties). Empty list when the
+        template has none. The proctoring cross-check path (seed params
+        per learner) reads through this.
+        """
+        ...
+
+    def close(self) -> None:
+        """Release DB connections. Store must not be used after close."""
+        ...
+
+
+def _sqlite_connect(dbapi_connection: sqlite3.Connection, _: object) -> None:
+    """Per-connection pragma setup (a-3). Mirrors telemetry/grading stores.
+
+    journal_mode=WAL  — readers never block the single writer.
+    synchronous=NORMAL — safe in WAL mode, avoids full fsync-per-commit.
+    busy_timeout=5000  — retry briefly under contention instead of
+                         `OperationalError: database is locked`.
+    """
+    cursor = dbapi_connection.cursor()
+    cursor.execute("PRAGMA journal_mode=WAL")
+    cursor.execute("PRAGMA synchronous=NORMAL")
+    cursor.execute("PRAGMA busy_timeout=5000")
+    cursor.close()
+
+
+def _as_utc(ts: datetime) -> datetime:
+    """Timestamps travel as naive datetime on SQLite, tz-aware elsewhere.
+
+    SQLite (via SQLModel) drops tzinfo; Postgres TIMESTAMP WITH TIME ZONE
+    keeps it. Normalizing on the read path makes the store's contract
+    tz-aware UTC regardless of the backend (D-027).
+    """
+    if ts.tzinfo is None:
+        return ts.replace(tzinfo=UTC)  # naive UTC read-side: label as UTC
+    return ts.astimezone(UTC)
+
+
+class SQLiteVariantStore:
+    """SQLite-backed VariantStore (SQLModel). Third protocol-wrapped store
+    of the D-027 family (first: SQLiteTraceStore, second: SQLiteGradeStore).
+    """
+
+    def __init__(self, db_path: Path | None = None) -> None:
+        self._db_path: Path = db_path if db_path is not None else Settings().db_path
+        self._engine = create_engine(f"sqlite:///{self._db_path}")
+        sa.event.listen(self._engine, "connect", _sqlite_connect)
+        SQLModel.metadata.create_all(self._engine)
+
+    @contextmanager
+    def _session(self) -> Iterator[Session]:
+        # expire_on_commit=False: identical session behavior to the other
+        # D-027 stores. save() never commits on the error path and the read
+        # paths never commit, but a uniform flag across the family keeps
+        # their detachment guarantees from diverging.
+        with Session(self._engine, expire_on_commit=False) as session:
+            yield session
+
+    def save(self, variant: VariantRecord) -> None:
+        # Plain INSERT, no merge: overwriting a stored variant would swap a
+        # learner's graded task underneath its trace and grade (audit
+        # corruption), so a duplicate identity is a race or bug to SURFACE,
+        # not paper over. The generator's cache path (get before generate)
+        # makes duplicate saves a programming error, not a normal flow.
+        # The trace store swallows its IntegrityError (dedup is the
+        # contract there); the grade store merges (latest-wins is the
+        # contract there); this store re-raises (first-wins is the
+        # contract here).
+        with self._session() as session:
+            try:
+                session.add(variant)
+                session.commit()
+            except sa.exc.IntegrityError:
+                session.rollback()
+                logger.debug(
+                    "variant insert rejected (identity already stored): "
+                    "learner=%s template=%s task=%s",
+                    variant.learner_id,
+                    variant.template_id,
+                    variant.task_id,
+                )
+                raise
+        logger.debug(
+            "variant saved: %s/%s task=%s seed=%s",
+            variant.learner_id,
+            variant.template_id,
+            variant.task_id,
+            variant.seed,
+        )
+
+    def get(self, learner_id: str, template_id: str) -> VariantRecord | None:
+        with self._session() as session:
+            record = session.get(VariantRecord, (learner_id, template_id))
+            if record is None:
+                return None
+            record.created_at = _as_utc(record.created_at)
+            # Detach from the session: callers must not depend on
+            # open-session ORM magic (lazy loads fail once it closes).
+            session.expunge(record)
+            return record
+
+    def get_by_task(self, task_id: str) -> VariantRecord | None:
+        with self._session() as session:
+            stmt = select(VariantRecord).where(VariantRecord.task_id == task_id)
+            record = session.exec(stmt).first()
+            if record is None:
+                return None
+            record.created_at = _as_utc(record.created_at)
+            session.expunge(record)
+            return record
+
+    def list_for_learner(self, learner_id: str) -> list[VariantRecord]:
+        with self._session() as session:
+            stmt = (
+                select(VariantRecord)
+                .where(VariantRecord.learner_id == learner_id)
+                # Chronological; task_id is a deterministic tie-break for
+                # variants stamped within the same instant.
+                .order_by(VariantRecord.created_at, VariantRecord.task_id)
+            )
+            results = session.exec(stmt).all()
+            for row in results:
+                row.created_at = _as_utc(row.created_at)
+                session.expunge(row)
+            return list(results)
+
+    def list_by_template(self, template_id: str) -> list[VariantRecord]:
+        with self._session() as session:
+            stmt = (
+                select(VariantRecord)
+                .where(VariantRecord.template_id == template_id)
+                # Chronological; learner_id is a deterministic tie-break.
+                .order_by(VariantRecord.created_at, VariantRecord.learner_id)
+            )
+            results = session.exec(stmt).all()
+            for row in results:
+                row.created_at = _as_utc(row.created_at)
+                session.expunge(row)
+            return list(results)
+
+    def close(self) -> None:
+        self._engine.dispose()

apps/ai-service/ai_service/variants/templates.py

@@ -0,0 +1,305 @@
+"""Task template library for seeded variant generation (D-029, REQ-3-005).
+
+A `TaskTemplate` binds a competency (D-021-aligned corpus ID), a statement
+skeleton with `{slot}` placeholders, typed `ParameterSlot`s, difficulty-
+normalization rubric anchors (the expected feature envelope that bounds
+variant fairness in the a-5 envelope test — grader-prompt shipment is the
+tracked P4 follow-up; grading is variant-blind today), and starter-file
+scaffolds served into the sandbox workdir (wired in P6).
+
+Slot sampling is PURE CODE: `random.Random(seed)` over typed slots — fully
+reproducible for a given seed, independent of the LLM. The LLM only renders
+the seeded slot values into the statement skeleton (D-020 defense).
+"""
+
+from __future__ import annotations
+
+import random
+import re
+from typing import Literal
+
+from pydantic import BaseModel, ConfigDict, Field, field_validator
+
+SlotType = Literal["enum", "int_range", "string_set"]
+
+
+class ParameterSlot(BaseModel):
+    """One typed fill-in for a statement skeleton."""
+
+    model_config = ConfigDict(frozen=True)
+
+    name: str = Field(min_length=1)
+    type: SlotType
+    values: list[str] = Field(default_factory=list)  # enum/string_set options
+    lo: int | None = None  # int_range bounds
+    hi: int | None = None
+
+    @field_validator("values")
+    @classmethod
+    def _values_nonempty_for_enums(cls, v: list[str], info) -> list[str]:
+        if info.data.get("type") in ("enum", "string_set") and not v:
+            raise ValueError(f"slot {info.data.get('name')!r} needs values")
+        return v
+
+    def sample(self, rng: random.Random) -> str | int:
+        """Deterministic sample from the seeded RNG. Validated after sampling."""
+        if self.type == "enum" or self.type == "string_set":
+            return rng.choice(self.values)
+        if self.type == "int_range":
+            lo = self.lo if self.lo is not None else 0
+            hi = self.hi if self.hi is not None else lo
+            if hi < lo:
+                raise ValueError(f"slot {self.name!r}: hi < lo")
+            return rng.randint(lo, hi)
+        raise ValueError(f"unsupported slot type: {self.type!r}")
+
+    def validate_value(self, value: str | int) -> bool:
+        """Is `value` schema-valid for this slot? (params JSON gate, a-5.)"""
+        if self.type in ("enum", "string_set"):
+            return isinstance(value, str) and value in self.values
+        if self.type == "int_range":
+            lo = self.lo if self.lo is not None else 0
+            hi = self.hi if self.hi is not None else lo
+            return isinstance(value, int) and lo <= value <= hi
+        return False
+
+
+class RubricAnchors(BaseModel):
+    """Difficulty-normalization anchors for the grader (a-5).
+
+    Expected FEATURE ENVELOPE (digest-space): the expected effort band
+    for this template, so two variants of one template are held to the
+    same bar regardless of which slot values a learner drew. The a-5
+    envelope test (tests/variants/test_generator.py) binds variants to
+    these bands in code, and — since Phase 4 (MH#4) — the grading engine
+    ships this envelope into the grader prompt
+    (grading/engine._anchors_context) and stamps the variant seed on the
+    GradeRecord, so the anchors gate variant fairness in BOTH tests and
+    the live rubric.
+    """
+
+    model_config = ConfigDict(frozen=True)
+
+    expected_edit_count_band: tuple[int, int]
+    expected_min_test_runs: int
+    expected_error_fix_cycles_band: tuple[int, int]
+    notes: str = ""
+
+
+class TaskTemplate(BaseModel):
+    """A reusable task shape; variants instantiate it per learner."""
+
+    model_config = ConfigDict(frozen=True)
+
+    id: str = Field(min_length=1)
+    competency_id: str = Field(min_length=1)  # D-021 corpus alignment
+    title: str
+    statement_skeleton: str = Field(min_length=1)  # {slot} placeholders
+    slots: list[ParameterSlot] = Field(min_length=1)
+    rubric_anchors: RubricAnchors
+    starter_files: dict[str, str] = Field(default_factory=dict)  # path -> content
+    test_command: str
+
+    @field_validator("statement_skeleton")
+    @classmethod
+    def _skeleton_placeholders(cls, v: str) -> str:
+        if "{" not in v or "}" not in v:
+            raise ValueError("statement_skeleton needs at least one {slot}")
+        return v
+
+    def render(self, params: dict[str, str | int]) -> str:
+        """Fill the skeleton with validated params."""
+        for slot in self.slots:
+            if slot.name not in params:
+                raise ValueError(f"missing param for slot {slot.name!r}")
+            if not slot.validate_value(params[slot.name]):
+                raise ValueError(f"invalid value for slot {slot.name!r}: {params[slot.name]!r}")
+        return self.statement_skeleton.format(**params)
+
+    def sample_params(self, seed: int) -> dict[str, str | int]:
+        """Seeded, reproducible, schema-valid slot values (pure code)."""
+        rng = random.Random(seed)
+        return {slot.name: slot.sample(rng) for slot in self.slots}
+
+
+# --- Template library (v0.3 initial set) --------------------------------------
+# Competency IDs are D-021-aligned with the Python corpus
+# (ai_service/corpus/learner_context.py) and the TS mock-data layer
+# (packages/mock-data/competency-stacks.ts: deterministic cid() scheme).
+
+TEMPLATES: dict[str, TaskTemplate] = {
+    "tpl-llm-judge": TaskTemplate(
+        id="tpl-llm-judge",
+        competency_id="stack-orchestration-c007",
+        title="Build an LLM-as-Judge Evaluator",
+        statement_skeleton=(
+            "Build a small LLM-as-judge evaluator for {domain} answers. "
+            "The judge must score each answer on {criterion} using a 0-4 scale, "
+            "return structured JSON, and handle at least {edge_cases} edge-case "
+            "answer classes (empty, off-topic, adversarial). Include a tiny "
+            "repro test set of at least {test_size} examples and print a summary "
+            "table of scores."
+        ),
+        slots=[
+            ParameterSlot(
+                name="domain",
+                type="enum",
+                values=["customer-support", "code-review", "summarization", "tutoring"],
+            ),
+            ParameterSlot(
+                name="criterion",
+                type="enum",
+                values=["factual-accuracy", "helpfulness", "safety", "completeness"],
+            ),
+            ParameterSlot(name="edge_cases", type="int_range", lo=2, hi=4),
+            ParameterSlot(name="test_size", type="int_range", lo=3, hi=8),
+        ],
+        rubric_anchors=RubricAnchors(
+            expected_edit_count_band=(3, 25),
+            expected_min_test_runs=2,
+            expected_error_fix_cycles_band=(0, 4),
+            notes="Slot draw changes the SCENARIO, not the engineering depth.",
+        ),
+        starter_files={
+            "README.md": (
+                "# LLM-as-Judge Evaluator\n\n"
+                "Implement `judge.py`:\n"
+                "- `score(answer: str) -> dict` — 0-4 on the named criterion\n"
+                "- structured JSON output (schema below)\n"
+                "- edge-case classes handled explicitly\n"
+                "- `pytest` must pass\n"
+            ),
+            "judge.py": "def score(answer: str) -> dict:\n    raise NotImplementedError\n",
+            "test_judge.py": "def test_placeholder():\n    assert True\n",
+        },
+        test_command="pytest -q",
+    ),
+    "tpl-guardrail-schema": TaskTemplate(
+        id="tpl-guardrail-schema",
+        competency_id="stack-orchestration-c008",
+        title="Schema Guardrail Pipeline",
+        statement_skeleton=(
+            "Implement an output-validation guardrail for a model returning "
+            "{entity} records. Validate against a typed schema with {field_count} "
+            "required fields, coerce or reject {failure_mode} failures, and emit "
+            "a fallback response for invalid payloads. Cover with at least "
+            "{test_size} unit tests including malformed JSON."
+        ),
+        slots=[
+            ParameterSlot(
+                name="entity",
+                type="enum",
+                values=["user-profile", "job-posting", "candidate", "invoice"],
+            ),
+            ParameterSlot(
+                name="failure_mode",
+                type="enum",
+                values=["strict-reject", "coerce-when-safe"],
+            ),
+            ParameterSlot(name="field_count", type="int_range", lo=4, hi=8),
+            ParameterSlot(name="test_size", type="int_range", lo=4, hi=10),
+        ],
+        rubric_anchors=RubricAnchors(
+            expected_edit_count_band=(3, 30),
+            expected_min_test_runs=2,
+            expected_error_fix_cycles_band=(0, 5),
+            notes="All slot draws land in the same engineering band.",
+        ),
+        starter_files={
+            "README.md": (
+                "# Schema Guardrail\n\nImplement `guardrail.py`:\n"
+                "- `validate(payload: dict) -> dict | Fallback`\n"
+                "- required-field checks, failure policy, fallback emission\n"
+            ),
+            "guardrail.py": "def validate(payload: dict):\n    raise NotImplementedError\n",
+            "test_guardrail.py": "def test_placeholder():\n    assert True\n",
+        },
+        test_command="pytest -q",
+    ),
+    "tpl-rag-chunker": TaskTemplate(
+        id="tpl-rag-chunker",
+        competency_id="stack-orchestration-c005",
+        title="RAG Chunking Strategy",
+        statement_skeleton=(
+            "Implement a document chunker for {doc_type} retrieval. Support "
+            "{strategy} chunking with a target size of ~{chunk_size} tokens, "
+            "preserve {invariant} across chunk boundaries, and evaluate overlap "
+            "quality with at least {test_size} fixture documents."
+        ),
+        slots=[
+            ParameterSlot(
+                name="doc_type",
+                type="enum",
+                values=["technical-docs", "legal-contracts", "transcripts"],
+            ),
+            ParameterSlot(
+                name="strategy",
+                type="enum",
+                values=["fixed-window", "semantic-boundary", "hybrid"],
+            ),
+            ParameterSlot(
+                name="invariant",
+                type="enum",
+                values=["code-block-integrity", "section-headers", "sentence-completeness"],
+            ),
+            ParameterSlot(name="chunk_size", type="int_range", lo=200, hi=800),
+            ParameterSlot(name="test_size", type="int_range", lo=3, hi=6),
+        ],
+        rubric_anchors=RubricAnchors(
+            expected_edit_count_band=(4, 35),
+            expected_min_test_runs=2,
+            expected_error_fix_cycles_band=(0, 6),
+            notes="Strategy draw changes implementation shape, not depth.",
+        ),
+        starter_files={
+            "README.md": (
+                "# RAG Chunker\n\nImplement `chunker.py`:\n"
+                "- `chunk(text: str) -> list[str]`\n- invariant preserved\n- tests green\n"
+            ),
+            "chunker.py": "def chunk(text: str) -> list[str]:\n    raise NotImplementedError\n",
+            "test_chunker.py": "def test_placeholder():\n    assert True\n",
+        },
+        test_command="pytest -q",
+    ),
+}
+
+
+_KNOWN_COMPETENCY_IDS: set[str] = {
+    # D-021: mirrored from ai_service/corpus/learner_context.py — the Python
+    # source of truth for stack-orchestration competencies used by v0.2 agents.
+    "stack-orchestration-c001",
+    "stack-orchestration-c002",
+    "stack-orchestration-c003",
+    "stack-orchestration-c004",
+    "stack-orchestration-c005",
+    "stack-orchestration-c007",
+    "stack-orchestration-c008",
+    "stack-orchestration-c011",
+    "stack-designer-c001",
+    "stack-designer-c002",
+    "stack-safety-c021",
+}
+
+
+def get_template(template_id: str) -> TaskTemplate | None:
+    return TEMPLATES.get(template_id)
+
+
+def template_for_competency(competency_id: str) -> list[TaskTemplate]:
+    return [t for t in TEMPLATES.values() if t.competency_id == competency_id]
+
+
+def validate_competency_binding() -> None:
+    """All templates must bind to known D-021 corpus competency IDs."""
+    for t in TEMPLATES.values():
+        if t.competency_id not in _KNOWN_COMPETENCY_IDS:
+            raise ValueError(
+                f"template {t.id!r} binds unknown competency {t.competency_id!r}"
+            )
+
+
+def slots_pattern_ok(skeleton: str, slots: list[ParameterSlot]) -> bool:
+    """Every {placeholder} in the skeleton has a matching slot and vice versa."""
+    placeholders = set(re.findall(r"\{([a-z_][a-z0-9_]*)\}", skeleton))
+    slot_names = {s.name for s in slots}
+    return placeholders == slot_names

apps/ai-service/ai_service/voice/__init__.py

@@ -0,0 +1,35 @@
+"""VoiceProvider protocol (D-030, REQ-3-006) — mirrors the LLMProvider seam.
+
+Two implementations in v0.3:
+- MockVoiceProvider — deterministic canned transcripts + canned tone WAV
+  chunks + scripted failure modes (tests + no-key default; tests NEVER call
+  a real voice API).
+- browser descriptor — not a provider but a FALLBACK HINT: the web client
+  selects browser-native SpeechRecognition/speechSynthesis when the server
+  reports no real voice backend.
+
+OpenAIAudioProvider (real server STT/TTS over OpenAI-compatible
+/audio/transcriptions + /audio/speech) is INTENTIONALLY NOT BUILT in v0.3 —
+deferred to v0.4 with KYC, when there is a real key and real users
+(GRILL CUT-1 / G-7). This protocol is its future drop-in seam.
+
+Boundary: `voice/` never imports `agents/` or `api/`.
+"""
+
+from ai_service.voice.base import (
+    TranscriptSegment,
+    VoiceDescriptor,
+    VoiceProvider,
+)
+from ai_service.voice.browser import BROWSER_FALLBACK_DESCRIPTOR
+from ai_service.voice.factory import voice_provider_from_settings
+from ai_service.voice.mock import MockVoiceProvider
+
+__all__ = [
+    "BROWSER_FALLBACK_DESCRIPTOR",
+    "MockVoiceProvider",
+    "TranscriptSegment",
+    "VoiceDescriptor",
+    "VoiceProvider",
+    "voice_provider_from_settings",
+]

apps/ai-service/ai_service/voice/base.py

@@ -0,0 +1,62 @@
+"""VoiceProvider protocol + shared voice contracts (D-030, REQ-3-006).
+
+Mirrors the LLMProvider seam (D-014 pattern): a narrow protocol the Examiner
+agent and the defense API compose via DI, with a deterministic mock and a
+browser-fallback descriptor. No network in this module — concrete providers
+live in their own modules and are selected by factory/config.
+"""
+
+from __future__ import annotations
+
+from collections.abc import AsyncIterator
+from typing import Literal, Protocol, runtime_checkable
+
+from pydantic import BaseModel, ConfigDict, Field
+
+VoiceRole = Literal["examiner", "learner"]
+
+
+class TranscriptSegment(BaseModel):
+    """One STT result: the transcribed text + timing metadata."""
+
+    model_config = ConfigDict(frozen=True)
+
+    text: str = Field(min_length=1)
+    language: str = "en"
+    duration_ms: int | None = None
+    confidence: float | None = Field(default=None, ge=0.0, le=1.0)
+
+
+class VoiceDescriptor(BaseModel):
+    """Capability descriptor served to the web client (D-030).
+
+    The assessment UI reads this to decide HOW the learner speaks/hears:
+    - `mode="server"`    → server-side STT/TTS (v0.4 real provider seam)
+    - `mode="browser"`   → browser-native SpeechRecognition/speechSynthesis
+    - `mode="mock"`      → deterministic no-op path (tests / no-key dev)
+    The descriptor never contains secrets — only capability hints.
+    """
+
+    model_config = ConfigDict(frozen=True)
+
+    mode: Literal["server", "browser", "mock"]
+    sr_available: bool
+    tts_available: bool
+    hint: str = ""
+
+
+@runtime_checkable
+class VoiceProvider(Protocol):
+    """The voice port (D-030): STT in, TTS out. Never imports agents/api."""
+
+    async def transcribe(self, audio: bytes, fmt: str) -> TranscriptSegment:
+        """STT: audio bytes (fmt: 'wav' | 'webm' | 'mp3') → transcript."""
+        ...
+
+    def synthesize(self, text: str, voice: str = "default") -> AsyncIterator[bytes]:
+        """TTS: text -> async byte chunks (audio stream).
+
+        Implementations may be async generators (async-def + yield) — the
+        consumer contract is `async for chunk in provider.synthesize(text)`.
+        """
+        ...

apps/ai-service/ai_service/voice/browser.py

@@ -0,0 +1,32 @@
+"""Browser-native fallback descriptor (D-030, CUT-1 / G-7, REQ-3-006).
+
+v0.3 has NO real server STT/TTS (deferred to v0.4 with KYC/keys — GRILL
+CUT-1). When the factory selects `browser` mode, the defense endpoints return
+this descriptor and the WEB CLIENT performs SpeechRecognition + speechSynthesis
+natively; the server persists text turns as usual.
+"""
+
+from __future__ import annotations
+
+from .base import VoiceDescriptor
+
+BROWSER_FALLBACK_DESCRIPTOR = VoiceDescriptor(
+    mode="browser",
+    sr_available=True,
+    tts_available=True,
+    hint=(
+        "No server voice backend configured. Use browser-native "
+        "SpeechRecognition for STT and speechSynthesis for TTS; send the "
+        "transcribed text to POST /v1/defense/{id}/answer ({text} form)."
+    ),
+)
+
+MOCK_DESCRIPTOR = VoiceDescriptor(
+    mode="mock",
+    sr_available=True,
+    tts_available=True,
+    hint=(
+        "Deterministic mock voice (tests / no-key dev). Server STT/TTS "
+        "endpoints serve canned responses; real server STT/TTS lands in v0.4."
+    ),
+)

apps/ai-service/ai_service/voice/defense_store.py

@@ -0,0 +1,506 @@
+"""DefenseStore — oral-defense persistence: protocol + SQLite impl (REQ-3-006, D-027).
+
+FOURTH protocol-wrapped store of the D-027 family and the first spanning
+TWO related tables: `defense_record` (the defense session + integrity
+signals) and `defense_turn` (the ordered examiner/learner transcript,
+FK → defense_record.id).
+
+Postgres-migration-ready (D-027): the protocol is the only surface the
+Examiner pipeline (task 5-2-01) and the defense endpoints (task 5-3-01)
+touch; swapping SQLiteDefenseStore for a Postgres implementation must
+not change call sites. Both tables use only portable column types
+(str / int / datetime / JSON), so the same SQLModel schema stands up
+unchanged on Postgres.
+
+Save semantics — where this sits among the D-027 stores (each has a
+deliberately different contract):
+    TraceStore.append   dedup-keep-first; IntegrityError SWALLOWED
+                        (at-least-once event ingest).
+    GradeStore.save     upsert-latest-wins (a regrade is latest-state).
+    VariantStore.save   insert-only first-wins; IntegrityError RAISED
+                        (reproducibility; a duplicate is a bug).
+    DefenseStore         a LIFECYCLE store:
+      start()           insert-only; a duplicate id raises
+                        (a defense id is minted once per session).
+      append_turn()      insert-only per (defense_id, seq); a duplicate
+                        seq raises AND an unknown defense_id raises (FK
+                        enforced) — a transcript turn must never silently
+                        vanish (it is the integrity/grading input) nor
+                        attach to a defense that does not exist.
+      finalize()         targeted UPDATE (status → finished; finished_at +
+                        integrity_signals JSON). Unknown id → None
+                        (documented below). Re-finalize overwrites
+                        signals + finished_at — latest-wins, mirroring
+                        GradeStore.save: a recomputed verdict replaces
+                        the previous one wholesale.
+
+append_turn does NOT police status (turns after finalize are a
+sequencing bug for the endpoints to prevent, task 5-3-01): the store
+enforces DATA integrity (FK + PK + non-empty), not workflow.
+
+integrity_signals (A-109): JSON dict on the record — long pauses,
+off-scope cadence markers and friends, computed by the Examiner over
+turn metadata and persisted by finalize for the Proctor/Mentor feed.
+An empty dict is legal (defense not finished, or a clean defense).
+
+Concurrency (a-3): WAL + synchronous=NORMAL + busy timeout at
+connection time (mirrors the other D-027 stores), PLUS foreign_keys=ON
+— this is the family's first real foreign key and it is actually
+enforced on SQLite, matching Postgres's native behavior (D-027 parity).
+
+`created_at` / `ts` contract: callers stamp UTC (datetime.now(UTC));
+SQLite stores them naive and read paths re-label tz-aware UTC (same
+boundary normalization as TelemetryEvent.ts / GradeRecord.created_at,
+so the contract holds on any backend).
+
+Boundary (D-027): `voice/` never imports `agents/` / `api/`; this
+module imports config only.
+"""
+
+import logging
+import sqlite3
+from collections.abc import Iterator
+from contextlib import contextmanager
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any, Literal, Protocol
+
+import sqlalchemy as sa
+from sqlalchemy import JSON, Index, String
+from sqlalchemy.orm import validates
+from sqlmodel import Field, Session, SQLModel, create_engine, select
+
+from ..config import Settings
+
+logger = logging.getLogger(__name__)
+
+DefenseStatus = Literal["in_progress", "finished"]
+_DEFENSE_STATUSES: frozenset[str] = frozenset(DefenseStatus.__args__)
+
+TurnRole = Literal["examiner", "learner"]
+_TURN_ROLES: frozenset[str] = frozenset(TurnRole.__args__)
+
+
+class DefenseRecord(SQLModel, table=True):
+    """A persisted oral-defense session; id is the PK.
+
+    Written by the defense endpoints (task 5-3-01) through the
+    DefenseStore protocol; read back by the endpoints, the Examiner
+    pipeline and the Proctor/Mentor feeds. Constraint enforcement
+    mirrors TelemetryEvent / GradeRecord / VariantRecord: sqlmodel
+    0.0.42's metaclass drops pydantic constraints on table models, so
+    SQLAlchemy `@validates` hooks enforce instead and the column types
+    stay Postgres-ready (D-027).
+
+    Field contract:
+        id                — non-empty defense identifier, minted once
+                            per session (a duplicate start raises).
+        learner_id        — non-empty learner identifier (same id space
+                            as traces, grades and variants).
+        task_id           — non-empty task identifier; the defense
+                            defends the submitted work for this trace
+                            key ((learner_id, task_id) joins to the
+                            trace/grade/variant the defense is about).
+        status            — in_progress | finished; the STORE owns the
+                            transition: start() forces in_progress,
+                            finalize() sets finished. Validated.
+        integrity_signals — A-109 signal dict (long pauses, off-scope
+                            cadence markers, ...); {} until finalize;
+                            JSON column. An empty dict is legal.
+        created_at        — UTC start timestamp.
+        finished_at       — UTC finalize timestamp; None while in
+                            progress.
+
+    `turns` (property): the seq-ordered DefenseTurn transcript, attached
+    ONLY by DefenseStore.get(); records from list_for_learner carry
+    turns == [] — call get() for a full transcript.
+    """
+
+    __tablename__ = "defense_record"
+    # The id PK covers point lookups; this secondary index covers
+    # list_for_learner ordered by created_at without a sort step
+    # (Postgres migration target D-027).
+    __table_args__ = (
+        Index("ix_defense_record_learner_created", "learner_id", "created_at"),
+    )
+
+    id: str = Field(primary_key=True)
+    learner_id: str
+    task_id: str
+    # Bare Literal annotations crash sqlmodel<=0.0.42's column inference
+    # (issubclass(TypeAlias, Enum)); an explicit sa_type + the validates
+    # hook below give the same contract: VARCHAR column, Literal-rejected
+    # values (same pattern as TelemetryEvent.kind).
+    status: DefenseStatus = Field(default="in_progress", sa_type=String)
+    # JSON column: stored as TEXT on SQLite, native JSONB on Postgres (D-027).
+    integrity_signals: dict[str, Any] = Field(default_factory=dict, sa_type=JSON)
+    created_at: datetime
+    finished_at: datetime | None = Field(default=None)
+
+    @property
+    def turns(self) -> list["DefenseTurn"]:
+        """Seq-ordered transcript; [] unless attached by get().
+
+        Table models reject ad-hoc attributes (pydantic __setattr__
+        raises on non-fields), so the store stashes the detached turn
+        list via object.__setattr__ and this read-only property surfaces
+        it. The returned list is a copy — caller mutations cannot
+        corrupt the stash.
+        """
+        return list(self.__dict__.get("_turns", []))
+
+    @validates("id", "learner_id", "task_id")
+    def _ids_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty identifier")
+        return value
+
+    @validates("status")
+    def _status_is_known(self, key: str, value: str) -> str:
+        if value not in _DEFENSE_STATUSES:
+            raise ValueError(f"unknown defense status: {value!r}")
+        return value
+
+
+class DefenseTurn(SQLModel, table=True):
+    """One examiner/learner dialogue turn; (defense_id, seq) is the PK.
+
+    Rows are append-only transcript entries written through
+    DefenseStore.append_turn. seq numbers the dialogue within one
+    defense starting at 0; monotonic assignment is the endpoints' job
+    (task 5-3-01), this model only rejects negatives — the same split
+    as TelemetryEvent.seq (model rejects < 0, store owns ordering).
+
+    Field contract:
+        defense_id  — non-empty; FK → defense_record.id. ENFORCED on
+                      SQLite via foreign_keys=ON (first real FK in the
+                      D-027 family; Postgres enforces FKs natively, so
+                      this keeps the backends equivalent, D-027).
+        seq         — turn index within the defense, >= 0. (defense_id,
+                      seq) is the PK: a duplicate raises instead of
+                      silently overwriting — the transcript is the
+                      integrity/grading input, a vanishing turn is
+                      audit corruption.
+        role        — examiner | learner (who spoke). Validated.
+        text        — non-empty utterance text (examiner question, or
+                      STT output for learner answers).
+        ts          — UTC utterance timestamp.
+        latency_ms  — per-turn pipeline latency in ms (STT + LLM TTFT +
+                      TTS, A-109); int or None. Populated by the
+                      endpoints (task 5-4-01); None allowed here — the
+                      store persists, it does not measure.
+        created_at  — UTC row-write timestamp.
+    """
+
+    __tablename__ = "defense_turn"
+    # The composite PK (defense_id, seq) doubles as the covering index
+    # for the per-defense seq-ordered read in get() — no secondary index
+    # needed (contrast defense_record's learner-listing index).
+
+    defense_id: str = Field(foreign_key="defense_record.id", primary_key=True)
+    seq: int = Field(primary_key=True)
+    role: TurnRole = Field(sa_type=String)
+    text: str
+    ts: datetime
+    latency_ms: int | None = Field(default=None)
+    created_at: datetime
+
+    @validates("defense_id")
+    def _defense_id_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty identifier")
+        return value
+
+    @validates("seq")
+    def _seq_non_negative(self, key: str, value: int) -> int:
+        if value < 0:
+            raise ValueError("seq must be >= 0 (ordering is the endpoints' job)")
+        return value
+
+    @validates("role")
+    def _role_is_known(self, key: str, value: str) -> str:
+        if value not in _TURN_ROLES:
+            raise ValueError(f"unknown turn role: {value!r}")
+        return value
+
+    @validates("text")
+    def _text_non_empty(self, key: str, value: str) -> str:
+        if not value:
+            raise ValueError(f"{key} must be a non-empty utterance string")
+        return value
+
+    @validates("latency_ms")
+    def _latency_non_negative(self, key: str, value: int | None) -> int | None:
+        # None is legal (not yet instrumented); a NEGATIVE latency is
+        # nonsense and surfaces as a construction error.
+        if value is not None and value < 0:
+            raise ValueError("latency_ms must be >= 0 or None")
+        return value
+
+
+class DefenseStore(Protocol):
+    """Persistence contract for oral-defense sessions + transcripts.
+
+    Implemented by SQLiteDefenseStore (v0.3, D-027); a Postgres
+    implementation must satisfy the same surface.
+    """
+
+    def start(self, defense: DefenseRecord) -> DefenseRecord:
+        """Insert a new defense. INSERT-ONLY: a duplicate id raises
+        sqlalchemy.exc.IntegrityError (a defense id is minted once per
+        session — surfacing, not swallowing, mirrors VariantStore).
+        The store owns the lifecycle: status is forced to "in_progress"
+        and finished_at to None, whatever the caller passed — only
+        finalize() may move a defense to finished. Returns the stored
+        record, detached from any DB session.
+        """
+        ...
+
+    def append_turn(self, defense_id: str, turn: DefenseTurn) -> DefenseTurn:
+        """Insert one transcript turn, ordered by (defense_id, seq).
+        turn.defense_id MUST equal the defense_id argument — a mismatch
+        raises ValueError (the defense identity must never be
+        ambiguous). A duplicate (defense_id, seq) raises
+        IntegrityError; an unknown defense_id raises IntegrityError
+        (FK enforced). Does NOT police status — sequencing turns vs
+        finalize is the endpoints' job (task 5-3-01). Returns the
+        stored turn, detached.
+        """
+        ...
+
+    def finalize(
+        self, defense_id: str, integrity_signals: dict[str, Any]
+    ) -> DefenseRecord | None:
+        """Seal the defense: status → "finished", finished_at = now(UTC),
+        integrity_signals stored as JSON. UNKNOWN defense_id → None
+        (documented choice: the API layer maps it to 404 without an
+        exception dance; contrast start/append_turn where IntegrityError
+        IS the contract — those are inserts, this is an update on a key
+        the caller may legitimately not hold). Re-finalize overwrites
+        signals + finished_at: latest-wins, mirroring GradeStore.save
+        (a recomputed verdict replaces the previous one wholesale).
+        Returns the updated record, detached, WITHOUT turns — get() is
+        the with-turns path.
+        """
+        ...
+
+    def get(self, defense_id: str) -> DefenseRecord | None:
+        """The defense with its FULL transcript (turns in seq order,
+        detached) and integrity signals; None when it does not exist.
+        Safe to pass across layers — no open-session ORM magic.
+        """
+        ...
+
+    def list_for_learner(self, learner_id: str) -> list[DefenseRecord]:
+        """All stored defenses for the learner, ordered by created_at
+        ascending (chronological; id breaks same-instant ties), WITHOUT
+        turns — records carry turns == []; call get() for a transcript.
+        Empty list when the learner has none.
+        """
+        ...
+
+    def close(self) -> None:
+        """Release DB connections. Store must not be used after close."""
+        ...
+
+
+def _sqlite_connect(dbapi_connection: sqlite3.Connection, _: object) -> None:
+    """Per-connection pragma setup (a-3). Mirrors the other D-027 stores.
+
+    journal_mode=WAL  — readers never block the single writer.
+    synchronous=NORMAL — safe in WAL mode, avoids full fsync-per-commit.
+    busy_timeout=5000  — retry briefly under contention instead of
+                         `OperationalError: database is locked`.
+    foreign_keys=ON    — NEW vs the family: defense_turn is the first
+                         real FK among the D-027 stores; SQLite leaves
+                         FKs OFF by default while Postgres enforces them
+                         natively, so the pragma keeps the backends
+                         equivalent (D-027 parity).
+    """
+    cursor = dbapi_connection.cursor()
+    cursor.execute("PRAGMA journal_mode=WAL")
+    cursor.execute("PRAGMA synchronous=NORMAL")
+    cursor.execute("PRAGMA busy_timeout=5000")
+    cursor.execute("PRAGMA foreign_keys=ON")
+    cursor.close()
+
+
+def _as_utc(ts: datetime) -> datetime:
+    """Timestamps travel as naive datetime on SQLite, tz-aware elsewhere.
+
+    SQLite (via SQLModel) drops tzinfo; Postgres TIMESTAMP WITH TIME ZONE
+    keeps it. Normalizing on the read path makes the store's contract
+    tz-aware UTC regardless of the backend (D-027).
+    """
+    if ts.tzinfo is None:
+        return ts.replace(tzinfo=UTC)  # naive UTC read-side: label as UTC
+    return ts.astimezone(UTC)
+
+
+class SQLiteDefenseStore:
+    """SQLite-backed DefenseStore (SQLModel). Fourth protocol-wrapped
+    store of the D-027 family (first: SQLiteTraceStore, second:
+    SQLiteGradeStore, third: SQLiteVariantStore) and the first spanning
+    two related tables.
+    """
+
+    def __init__(self, db_path: Path | None = None) -> None:
+        self._db_path: Path = db_path if db_path is not None else Settings().db_path
+        self._engine = create_engine(f"sqlite:///{self._db_path}")
+        sa.event.listen(self._engine, "connect", _sqlite_connect)
+        SQLModel.metadata.create_all(self._engine)
+
+    @contextmanager
+    def _session(self) -> Iterator[Session]:
+        # expire_on_commit=False: identical session behavior to the other
+        # D-027 stores. start/append_turn return the caller's instance
+        # after commit and get/finalize return rows expunged mid-session;
+        # a uniform flag across the family keeps their detachment
+        # guarantees from diverging.
+        with Session(self._engine, expire_on_commit=False) as session:
+            yield session
+
+    def start(self, defense: DefenseRecord) -> DefenseRecord:
+        # The store owns the lifecycle: a defense is BORN in_progress and
+        # only finalize() may move it to finished. A smuggled "finished"
+        # status is normalized away, not rejected — the insert itself
+        # stays insert-only, and a duplicate id raises to the caller
+        # (mirroring VariantStore: the id is minted once per session).
+        defense.status = "in_progress"
+        defense.finished_at = None
+        with self._session() as session:
+            try:
+                session.add(defense)
+                session.commit()
+            except sa.exc.IntegrityError:
+                session.rollback()
+                logger.debug("defense start rejected (id already stored): %s", defense.id)
+                raise
+        logger.debug(
+            "defense started: %s learner=%s task=%s",
+            defense.id,
+            defense.learner_id,
+            defense.task_id,
+        )
+        return defense
+
+    def append_turn(self, defense_id: str, turn: DefenseTurn) -> DefenseTurn:
+        # The explicit defense_id argument is the defense identity for
+        # this write; a turn object claiming another defense is a
+        # programming error — surface it before touching the DB.
+        if turn.defense_id != defense_id:
+            raise ValueError(
+                f"turn.defense_id {turn.defense_id!r} does not match the "
+                f"defense_id argument {defense_id!r}"
+            )
+        with self._session() as session:
+            try:
+                session.add(turn)
+                session.commit()
+            except sa.exc.IntegrityError:
+                # Two possible causes, both surfaced, neither swallowed:
+                # duplicate (defense_id, seq) PK — a transcript turn must
+                # never silently vanish; unknown defense_id — the FK
+                # (foreign_keys=ON) rejects the orphan.
+                session.rollback()
+                logger.debug(
+                    "defense turn rejected (duplicate (defense_id, seq) "
+                    "or unknown defense_id): defense=%s seq=%s",
+                    defense_id,
+                    turn.seq,
+                )
+                raise
+        logger.debug(
+            "defense turn appended: %s seq=%d role=%s",
+            defense_id,
+            turn.seq,
+            turn.role,
+        )
+        return turn
+
+    def finalize(
+        self, defense_id: str, integrity_signals: dict[str, Any]
+    ) -> DefenseRecord | None:
+        # A None signals blob would break the read contract (signals are
+        # a dict, {} until finalize); reject before writing.
+        if not isinstance(integrity_signals, dict):
+            raise ValueError(
+                "integrity_signals must be a JSON-object dict, got "
+                f"{type(integrity_signals).__name__}"
+            )
+        with self._session() as session:
+            record = session.get(DefenseRecord, defense_id)
+            if record is None:
+                # Documented unknown-id behavior: None, not a raise — the
+                # defense endpoints map this to 404. Contrast start() /
+                # append_turn(), where IntegrityError IS the contract.
+                return None
+            # Latest-wins re-finalize, mirroring GradeStore.save: a
+            # recomputed verdict (fresh signals) replaces the stored one
+            # wholesale; status just stays finished.
+            record.status = "finished"
+            record.finished_at = datetime.now(UTC)
+            record.integrity_signals = integrity_signals
+            session.commit()
+            record.created_at = _as_utc(record.created_at)
+            if record.finished_at is not None:
+                record.finished_at = _as_utc(record.finished_at)
+            # Detach from the session: callers must not depend on
+            # open-session ORM magic (lazy loads fail once it closes).
+            session.expunge(record)
+        logger.debug(
+            "defense finalized: %s signals=%s", defense_id, sorted(integrity_signals)
+        )
+        return record
+
+    def get(self, defense_id: str) -> DefenseRecord | None:
+        with self._session() as session:
+            record = session.get(DefenseRecord, defense_id)
+            if record is None:
+                return None
+            record.created_at = _as_utc(record.created_at)
+            if record.finished_at is not None:
+                record.finished_at = _as_utc(record.finished_at)
+            stmt = (
+                select(DefenseTurn)
+                .where(DefenseTurn.defense_id == defense_id)
+                .order_by(DefenseTurn.seq)
+            )
+            turns = session.exec(stmt).all()
+            for turn in turns:
+                turn.ts = _as_utc(turn.ts)
+                turn.created_at = _as_utc(turn.created_at)
+                # Detach each turn: the transcript must be usable once
+                # the session closes (no lazy-load magic).
+                session.expunge(turn)
+            session.expunge(record)
+            # Table models reject ad-hoc attributes (pydantic __setattr__
+            # raises on non-fields), so the seq-ordered transcript is
+            # stashed via object.__setattr__ and surfaced through the
+            # read-only `turns` property. Rows are detached either way —
+            # safe to pass across layers.
+            object.__setattr__(record, "_turns", list(turns))
+            return record
+
+    def list_for_learner(self, learner_id: str) -> list[DefenseRecord]:
+        with self._session() as session:
+            stmt = (
+                select(DefenseRecord)
+                .where(DefenseRecord.learner_id == learner_id)
+                # Chronological; id is a deterministic tie-break for
+                # defenses stamped within the same instant.
+                .order_by(DefenseRecord.created_at, DefenseRecord.id)
+            )
+            results = session.exec(stmt).all()
+            for row in results:
+                row.created_at = _as_utc(row.created_at)
+                if row.finished_at is not None:
+                    row.finished_at = _as_utc(row.finished_at)
+                # Turns are deliberately NOT loaded here: the list feed
+                # (Proctor/Mentor) needs session headers, not full
+                # transcripts — get() is the with-turns path.
+                session.expunge(row)
+            return list(results)
+
+    def close(self) -> None:
+        self._engine.dispose()

apps/ai-service/ai_service/voice/factory.py

@@ -0,0 +1,37 @@
+"""Voice provider factory (D-030, REQ-3-006).
+
+`AI_VOICE_PROVIDER = browser | mock` (default: mock — the no-key path is
+first-class). The real server provider (`openai-audio`) is a v0.4 seam and
+is REJECTED here with a clear error naming the deferral, so a stale env var
+can't silently pretend a real backend exists.
+"""
+
+from __future__ import annotations
+
+from ..config import Settings
+from .base import VoiceProvider
+from .mock import MockVoiceProvider
+
+
+class UnknownVoiceProviderError(ValueError):
+    """Raised for a provider name outside the v0.3 contract."""
+
+
+def voice_provider_from_settings(settings: Settings) -> VoiceProvider:
+    """Select the voice provider by settings (env `AI_VOICE_PROVIDER`)."""
+    name = (settings.voice_provider or "mock").strip().lower()
+    if name == "mock":
+        return MockVoiceProvider()
+    if name == "browser":
+        # Browser mode is a CLIENT-side capability: the server composes the
+        # same MockVoiceProvider (typed fallback answers still work; the UI
+        # uses the descriptor for mic/speech). See browser.py.
+        return MockVoiceProvider()
+    if name in ("openai-audio", "openai", "server"):
+        raise UnknownVoiceProviderError(
+            "real server STT/TTS (OpenAIAudioProvider) is deferred to v0.4 "
+            "(GRILL CUT-1 / G-7): set AI_VOICE_PROVIDER=mock or browser"
+        )
+    raise UnknownVoiceProviderError(
+        f"unknown AI_VOICE_PROVIDER {name!r}: use 'mock' or 'browser'"
+    )

apps/ai-service/ai_service/voice/mock.py

@@ -0,0 +1,96 @@
+"""Deterministic MockVoiceProvider (D-030, REQ-3-006).
+
+Canned transcripts (scripted per test via queue) + canned 1kHz-tone WAV bytes
++ scripted failure modes. Two identical transcribe calls yield identical
+segments; tests NEVER touch a real voice API (conftest cloud-free rule).
+"""
+
+from __future__ import annotations
+
+import asyncio
+import io
+import math
+import struct
+import wave
+from collections.abc import AsyncIterator
+
+from .base import TranscriptSegment
+
+
+def _tone_wav(duration_ms: int = 250, freq_hz: float = 1000.0) -> bytes:
+    """A small, deterministic 16-bit mono WAV: a sine tone (stdlib only)."""
+    rate = 8000
+    n_samples = max(1, int(rate * duration_ms / 1000))
+    buf = io.BytesIO()
+    with wave.open(buf, "wb") as w:
+        w.setnchannels(1)
+        w.setsampwidth(2)
+        w.setframerate(rate)
+        for i in range(n_samples):
+            sample = int(12000 * math.sin(2 * math.pi * freq_hz * i / rate))
+            w.writeframes(struct.pack("<h", sample))
+    return buf.getvalue()
+
+
+class MockVoiceFailure(RuntimeError):
+    """Scripted failure mode for tests."""
+
+
+class MockVoiceProvider:
+    """Deterministic voice provider: scripted STT, canned-tone TTS.
+
+    - `transcribe`: pops the next scripted transcript from a queue (or a
+      default); two identical calls with the same queue state are identical.
+      Failure mode: raise MockVoiceFailure when the queue holds a failure
+      marker (the string "FAIL") or `audio` is empty.
+    - `synthesize`: yields the canned tone WAV in fixed-size chunks; failure
+      mode: empty text raises MockVoiceFailure.
+    """
+
+    def __init__(self, transcripts: list[str] | None = None) -> None:
+        self._transcripts = list(transcripts or [])
+        self._cursor = 0
+        self.transcribe_calls = 0
+        self.synthesize_calls = 0
+
+    def script(self, transcripts: list[str]) -> None:
+        """Replace the scripted queue (tests set expectations up front)."""
+        self._transcripts = list(transcripts)
+        self._cursor = 0
+
+    async def transcribe(self, audio: bytes, fmt: str) -> TranscriptSegment:
+        self.transcribe_calls += 1
+        if not audio:
+            raise MockVoiceFailure("no audio bytes provided")
+        if not self._transcripts:
+            raise MockVoiceFailure("transcript queue exhausted — script it")
+        item = self._transcripts[self._cursor]
+        self._cursor = (self._cursor + 1) % len(self._transcripts)
+        if item == "FAIL":
+            raise MockVoiceFailure("scripted STT failure")
+        return TranscriptSegment(
+            text=item,
+            duration_ms=max(1, len(audio) // 32),  # deterministic pseudo-duration
+        )
+
+    async def synthesize(self, text: str, voice: str = "default") -> AsyncIterator[bytes]:  # noqa: ASYNC109 (protocol parity)
+        # NOTE: protocol parity matters more than the async-generator purity
+        # lint; the real provider seam (v0.4) will stream over HTTP.
+        self.synthesize_calls += 1
+        if not text:
+            raise MockVoiceFailure("cannot synthesize empty text")
+        wav = _tone_wav(duration_ms=min(2000, max(120, len(text) * 12)))
+        for i in range(0, len(wav), 1024):
+            yield wav[i : i + 1024]
+            await asyncio.sleep(0)  # yield to the loop like a network stream
+
+
+# Protocol-shape parity guard (mock must satisfy the D-030 port).
+from .base import VoiceProvider  # noqa: E402
+
+
+def _assert_protocol() -> None:
+    assert isinstance(MockVoiceProvider(), VoiceProvider)
+
+
+_assert_protocol()

apps/ai-service/pyproject.toml

@@ -18,6 +18,9 @@ dependencies = [
     "sqlalchemy>=2.0,<2.1",
     "websockets>=13,<16",
     "aiofiles>=24.1,<26",
+    # POST /v1/defense/{id}/answer multipart audio (REQ-3-006): FastAPI
+    # form/File parsing requires python-multipart at runtime.
+    "python-multipart>=0.0.32,<0.1",
 ]
 
 [project.optional-dependencies]

apps/ai-service/scripts/bootstrap.sh

@@ -1,28 +1,73 @@
 #!/usr/bin/env bash
 # Idempotent bootstrap: create venv + install deps.
-# Handles Debian systems without python3-venv/ensurepip via --without-pip + get-pip.
+# Handles Debian/Ubuntu systems without python3-venv/ensurepip via --without-pip + get-pip.
+# v2 (v0.3.5): recovers from a poisoned partial .venv left by a failed earlier
+# attempt, cleans before each retry, and dies with a distro-specific fix hint
+# when venv creation is impossible (e.g. missing python3.XX-venv package).
 set -euo pipefail
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
 APP_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
 VENV="$APP_DIR/.venv"
 
+venv_usable() {
+  [ -x "$VENV/bin/python3" ]
+}
+
+rm_broken_venv() {
+  echo "bootstrap: removing broken partial .venv from a failed earlier attempt" >&2
+  rm -rf "$VENV"
+}
+
 mkdir -p "$HOME/.cache/ciagent"
 
-if [ ! -x "$VENV/bin/python3" ]; then
-  if python3 -m venv "$VENV" 2>/dev/null; then
+if venv_usable && [ ! -x "$VENV/bin/pip" ]; then
+  # A usable python3 without pip means the --without-pip fallback half-ran and
+  # the get-pip step never completed: start over cleanly.
+  rm_broken_venv
+fi
+
+if ! venv_usable; then
+  if [ -d "$VENV" ]; then
+    # Directory exists but no working python3: remains of a crashed venv create.
+    rm_broken_venv
+  fi
+  if python3 -m venv "$VENV" 2>/tmp/venv-create.err; then
     :
   else
-    # No ensurepip available — create bare venv and bootstrap pip separately.
-    python3 -m venv --without-pip "$VENV"
+    rm -rf "$VENV"
+    if python3 -m venv --without-pip "$VENV" 2>>/tmp/venv-create.err; then
+      :
+    else
+      rm -rf "$VENV"
+      PYVER="$(python3 -c 'import sys; print("%d.%d" % sys.version_info[:2])' 2>/dev/null || true)"
+      PKG="python3-venv"
+      [ -n "$PYVER" ] && PKG="python${PYVER}-venv"
+      echo "bootstrap: could not create a virtual environment." >&2
+      echo "  python3 reported:" >&2
+      sed 's/^/    /' /tmp/venv-create.err >&2 || true
+      echo "  fix (Debian/Ubuntu): install the venv support package, then re-run nextcraft bootstrap:" >&2
+      echo "    apt install $PKG" >&2
+      exit 1
+    fi
   fi
 fi
 
 if [ ! -x "$VENV/bin/pip" ]; then
   GET_PIP="$HOME/.cache/ciagent/get-pip.py"
   if [ ! -f "$GET_PIP" ]; then
-    curl -sSf --max-time 60 https://bootstrap.pypa.io/get-pip.py -o "$GET_PIP"
+    if ! curl -sSf --max-time 60 https://bootstrap.pypa.io/get-pip.py -o "$GET_PIP"; then
+      rm -rf "$VENV"
+      echo "bootstrap: get-pip.py download failed (no network?)." >&2
+      echo "  fix: restore network access and re-run nextcraft bootstrap" >&2
+      exit 1
+    fi
+  fi
+  if ! "$VENV/bin/python3" "$GET_PIP" --quiet; then
+    rm -rf "$VENV"
+    echo "bootstrap: pip installation into the venv failed." >&2
+    echo "  fix: re-run nextcraft bootstrap (the venv was cleaned; this retry is safe)" >&2
+    exit 1
   fi
-  "$VENV/bin/python3" "$GET_PIP" --quiet
 fi
 
 "$VENV/bin/pip" install --quiet --upgrade pip

apps/ai-service/scripts/dev.sh

@@ -1,5 +1,7 @@
 #!/usr/bin/env bash
-# Dev server: export secrets (if present) then run uvicorn on :8420.
+# Dev server: export secrets (if present) then run uvicorn.
+# Binds 0.0.0.0 by default so the stack is reachable from other machines
+# (v0.3.5 network mode) — set AI_HOST=127.0.0.1 in .env to revert to loopback.
 set -euo pipefail
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
 APP_DIR="$(cd "$SCRIPT_DIR/.." && pwd)"
@@ -7,7 +9,7 @@ REPO_ROOT="$(cd "$APP_DIR/../.." && pwd)"
 VENV="$APP_DIR/.venv"
 
 if [ ! -x "$VENV/bin/uvicorn" ]; then
-  echo "venv missing — run scripts/bootstrap.sh first" >&2
+  echo "venv missing — run nextcraft bootstrap first (or: bash scripts/bootstrap.sh)" >&2
   exit 1
 fi
 
@@ -22,5 +24,17 @@ if [ -f "$SECRETS" ]; then
   done < "$SECRETS"
 fi
 
+ENV_FILE="$APP_DIR/.env"
+if [ -f "$ENV_FILE" ]; then
+  while IFS='=' read -r key value; do
+    case "$key" in
+      AI_HOST|AI_PORT|AI_CORS_ORIGINS) export "$key=$value" ;;
+    esac
+  done < "$ENV_FILE"
+fi
+
+HOST="${AI_HOST:-0.0.0.0}"
+PORT="${AI_PORT:-8420}"
+
 cd "$APP_DIR"
-exec "$VENV/bin/uvicorn" ai_service.main:app --reload --port 8420
+exec "$VENV/bin/uvicorn" ai_service.main:app --reload --host "$HOST" --port "$PORT"

apps/ai-service/scripts/sandbox-agent.py

@@ -0,0 +1,691 @@
+#!/usr/bin/env python3
+"""sandbox-agent — stdlib-only in-sandbox telemetry capture agent (REQ-3-003).
+
+D-031: this file is copied into the sandbox namespace and runs against the
+system Python — no third-party packages are importable there, so this module
+depends on the standard library ONLY (the test suite enforces this with an
+AST scan of the file).
+
+What it does:
+  * wraps a non-interactive `/bin/sh` REPL: each stdin line is executed via
+    `sh -c` inside the workspace and reported as `stdin` -> `command` ->
+    `stdout` -> `run_result`/`test_result` events;
+  * polls the workspace tree (~250 ms) and emits `file_diff` events
+    (created/modified/deleted with unified diffs) plus periodic `activity`
+    heartbeats;
+  * streams events to ai-service as TelemetryEvent-shaped JSON frames over a
+    raw-socket RFC 6455 WebSocket client (no `websockets` package exists in
+    the namespace — the client handshake + frame codec is implemented here);
+  * at-least-once delivery (D-026): every event is appended to an fsync'd
+    JSONL spool file inside the workdir BEFORE any send attempt; on
+    disconnect the spool grows; after reconnect (exponential backoff) the
+    spool is flushed oldest-first. The server dedups on (learner, task, seq)
+    so replayed duplicates are harmless — loss is not tolerated.
+
+Configured entirely through env baked at spawn time:
+  NC_LEARNER_ID / NC_TASK_ID / NC_INGEST_URL / NC_SANDBOX_ID   (required)
+  NC_WORKSPACE            workspace root to watch/run in (default: cwd)
+  NC_SPOOL                spool path (default: <workspace>/.nc-agent/spool.jsonl)
+  NC_POLL_INTERVAL_S / NC_ACTIVITY_INTERVAL_S / NC_COMMAND_TIMEOUT_S
+  NC_BACKOFF_BASE_S / NC_BACKOFF_MAX_S                          (optional knobs)
+
+Sequencing survives process restarts (incl. SIGKILL): on boot the spool is
+replayed into the pending queue and `seq` resumes at max(spooled seq) + 1.
+"""
+
+from __future__ import annotations
+
+import base64
+import difflib
+import hashlib
+import json
+import os
+import secrets
+import socket
+import ssl
+import struct
+import subprocess
+import sys
+import threading
+import time
+import urllib.parse
+from collections import deque
+from collections.abc import Mapping
+from dataclasses import dataclass
+from datetime import UTC, datetime
+from pathlib import Path
+from typing import Any
+
+_WS_GUID = "258EAFA5-E914-47DA-95CA-C5AB0DC85B11"
+_EVENT_KINDS = frozenset(
+    {"command", "file_diff", "run_result", "test_result", "activity", "stdin", "stdout"}
+)
+_AGENT_DIR_PREFIX = ".nc-"  # agent-private paths (spool) are excluded from watching
+_MAX_DIFF_BYTES = 64 * 1024  # files larger than this are reported truncated, no diff
+_MAX_OUTPUT_CHARS = 64 * 1024  # captured stdout/stderr tail cap per command
+_HANDSHAKE_MAX_BYTES = 64 * 1024
+
+
+# --------------------------------------------------------------------------- config
+
+
+@dataclass(frozen=True)
+class AgentConfig:
+    """Runtime configuration, normally built from `NC_*` env baked at spawn."""
+
+    learner_id: str
+    task_id: str
+    ingest_url: str
+    sandbox_id: str
+    workspace: Path
+    spool_path: Path
+    poll_interval_s: float = 0.25
+    activity_interval_s: float = 5.0
+    command_timeout_s: float = 30.0
+    backoff_base_s: float = 0.25
+    backoff_max_s: float = 8.0
+
+    def __post_init__(self) -> None:
+        for name in ("learner_id", "task_id", "ingest_url", "sandbox_id"):
+            if not getattr(self, name):
+                raise ValueError(f"missing required config: NC_{name.upper()}")
+
+    @classmethod
+    def from_env(cls, env: Mapping[str, str] | None = None) -> AgentConfig:
+        src = os.environ if env is None else env
+        workspace = Path(src.get("NC_WORKSPACE") or os.getcwd()).resolve()
+        return cls(
+            learner_id=src.get("NC_LEARNER_ID", ""),
+            task_id=src.get("NC_TASK_ID", ""),
+            ingest_url=src.get("NC_INGEST_URL", ""),
+            sandbox_id=src.get("NC_SANDBOX_ID", ""),
+            workspace=workspace,
+            spool_path=Path(
+                src.get("NC_SPOOL") or (workspace / ".nc-agent" / "spool.jsonl")
+            ),
+            poll_interval_s=float(src.get("NC_POLL_INTERVAL_S", "0.25")),
+            activity_interval_s=float(src.get("NC_ACTIVITY_INTERVAL_S", "5.0")),
+            command_timeout_s=float(src.get("NC_COMMAND_TIMEOUT_S", "30.0")),
+            backoff_base_s=float(src.get("NC_BACKOFF_BASE_S", "0.25")),
+            backoff_max_s=float(src.get("NC_BACKOFF_MAX_S", "8.0")),
+        )
+
+
+# --------------------------------------------------------------------------- spool
+
+
+class Spool:
+    """Append-only JSONL spool with per-append fsync (survives SIGKILL).
+
+    `rewrite` swaps in a compacted file atomically (tmp file + os.replace).
+    Lines are stored without trailing newlines in memory, one per line on disk.
+    """
+
+    def __init__(self, path: Path) -> None:
+        self._path = path
+        path.parent.mkdir(parents=True, exist_ok=True)
+
+    @property
+    def path(self) -> Path:
+        return self._path
+
+    def append(self, line: str) -> None:
+        with self._path.open("a", encoding="utf-8") as fh:
+            fh.write(line + "\n")
+            fh.flush()
+            os.fsync(fh.fileno())
+
+    def read_all(self) -> list[str]:
+        if not self._path.exists():
+            return []
+        with self._path.open("r", encoding="utf-8") as fh:
+            return [line.rstrip("\n") for line in fh if line.strip()]
+
+    def rewrite(self, lines: list[str]) -> None:
+        tmp = self._path.with_name(self._path.name + ".tmp")
+        with tmp.open("w", encoding="utf-8") as fh:
+            for line in lines:
+                fh.write(line + "\n")
+            fh.flush()
+            os.fsync(fh.fileno())
+        os.replace(tmp, self._path)
+
+
+# --------------------------------------------------------------- websocket codec
+
+
+def _encode_frame(opcode: int, payload: bytes) -> bytes:
+    """RFC 6455 client frame: FIN set, always masked (servers require it)."""
+    header = bytearray([0x80 | opcode])
+    n = len(payload)
+    if n < 126:
+        header.append(0x80 | n)
+    elif n < 65536:
+        header.append(0x80 | 126)
+        header += struct.pack("!H", n)
+    else:
+        header.append(0x80 | 127)
+        header += struct.pack("!Q", n)
+    mask = secrets.token_bytes(4)
+    header += mask
+    masked = bytes(b ^ mask[i % 4] for i, b in enumerate(payload))
+    return bytes(header) + masked
+
+
+class WsConnection:
+    """Minimal blocking RFC 6455 client over a raw socket (stdlib only)."""
+
+    def __init__(self, sock: socket.socket) -> None:
+        self._sock = sock
+        self._write_lock = threading.Lock()
+
+    @classmethod
+    def connect(cls, url: str, timeout_s: float = 5.0) -> WsConnection:
+        parts = urllib.parse.urlsplit(url)
+        if parts.scheme not in ("ws", "wss"):
+            raise ValueError(f"unsupported scheme in NC_INGEST_URL: {parts.scheme!r}")
+        host = parts.hostname or "localhost"
+        port = parts.port or (443 if parts.scheme == "wss" else 80)
+        path = parts.path or "/"
+        if parts.query:
+            path += "?" + parts.query
+
+        sock = socket.create_connection((host, port), timeout=timeout_s)
+        if parts.scheme == "wss":
+            sock = ssl.create_default_context().wrap_socket(sock, server_hostname=host)
+
+        key = base64.b64encode(secrets.token_bytes(16)).decode("ascii")
+        request = (
+            f"GET {path} HTTP/1.1\r\n"
+            f"Host: {host}:{port}\r\n"
+            "Upgrade: websocket\r\n"
+            "Connection: Upgrade\r\n"
+            f"Sec-WebSocket-Key: {key}\r\n"
+            "Sec-WebSocket-Version: 13\r\n\r\n"
+        )
+        sock.sendall(request.encode("ascii"))
+        response = cls._read_http_response(sock)
+        cls._validate_handshake(response, key)
+        return cls(sock)
+
+    @staticmethod
+    def _read_http_response(sock: socket.socket) -> bytes:
+        buf = b""
+        while b"\r\n\r\n" not in buf:
+            chunk = sock.recv(4096)
+            if not chunk:
+                raise ConnectionError("server closed during WebSocket handshake")
+            buf += chunk
+            if len(buf) > _HANDSHAKE_MAX_BYTES:
+                raise ConnectionError("handshake response exceeded size cap")
+        return buf.split(b"\r\n\r\n", 1)[0]
+
+    @staticmethod
+    def _validate_handshake(response: bytes, key: str) -> None:
+        head = response.decode("latin-1")
+        lines = head.split("\r\n")
+        if not lines or " 101" not in lines[0]:
+            raise ConnectionError(f"handshake rejected: {lines[0] if lines else '<empty>'}")
+        headers = {}
+        for line in lines[1:]:
+            if ":" in line:
+                name, _, value = line.partition(":")
+                headers[name.strip().lower()] = value.strip()
+        expect = base64.b64encode(
+            hashlib.sha1((key + _WS_GUID).encode("ascii")).digest()
+        ).decode("ascii")
+        if headers.get("sec-websocket-accept") != expect:
+            raise ConnectionError("bad Sec-WebSocket-Accept in handshake response")
+
+    # -- send ------------------------------------------------------------
+    def send_text(self, text: str) -> None:
+        with self._write_lock:
+            self._sock.sendall(_encode_frame(0x1, text.encode("utf-8")))
+
+    def _send_frame(self, opcode: int, payload: bytes) -> None:
+        with self._write_lock:
+            self._sock.sendall(_encode_frame(opcode, payload))
+
+    # -- receive ---------------------------------------------------------
+    def recv_message(self, timeout_s: float) -> tuple[int, bytes] | None:
+        """Return (opcode, payload) for a data/close frame, or None on timeout.
+
+        Ping frames are answered with pong internally and never surfaced;
+        pongs are swallowed. Fragmented messages are reassembled. Raises
+        ConnectionError/OSError when the socket breaks.
+        """
+        deadline = time.monotonic() + timeout_s
+        fragments = bytearray()
+        frag_opcode = 0
+        while True:
+            frame = self._recv_one_frame(deadline)
+            if frame is None:
+                return None
+            fin, opcode, payload = frame
+            if opcode == 0x9:  # ping
+                self._send_frame(0xA, payload)
+                continue
+            if opcode == 0xA:  # pong
+                continue
+            if opcode == 0x0:  # continuation
+                fragments += payload
+            else:
+                fragments = bytearray(payload)
+                frag_opcode = opcode
+            if fin:
+                return frag_opcode, bytes(fragments)
+
+    def _recv_one_frame(self, deadline: float) -> tuple[bool, int, bytes] | None:
+        header = self._read_exact(2, deadline)
+        if header is None:
+            return None
+        b0, b1 = header[0], header[1]
+        fin = bool(b0 & 0x80)
+        opcode = b0 & 0x0F
+        length = b1 & 0x7F
+        if length == 126:
+            ext = self._read_exact(2, deadline)
+            if ext is None:
+                return None
+            length = struct.unpack("!H", ext)[0]
+        elif length == 127:
+            ext = self._read_exact(8, deadline)
+            if ext is None:
+                return None
+            length = struct.unpack("!Q", ext)[0]
+        mask = self._read_exact(4, deadline) if (b1 & 0x80) else b""
+        if mask is None:
+            return None
+        payload = self._read_exact(length, deadline) if length else b""
+        if payload is None:
+            return None
+        if mask:
+            payload = bytes(b ^ mask[i % 4] for i, b in enumerate(payload))
+        return fin, opcode, payload
+
+    def _read_exact(self, n: int, deadline: float) -> bytes | None:
+        buf = bytearray()
+        while len(buf) < n:
+            remaining = deadline - time.monotonic()
+            if remaining <= 0:
+                return None
+            self._sock.settimeout(remaining)
+            try:
+                chunk = self._sock.recv(n - len(buf))
+            except TimeoutError:
+                return None
+            if not chunk:
+                raise ConnectionError("peer closed the WebSocket connection")
+            buf += chunk
+        return bytes(buf)
+
+    def close(self) -> None:
+        try:
+            self._sock.close()
+        except OSError:
+            pass
+
+
+# --------------------------------------------------------------------------- agent
+
+
+class Agent:
+    """Wires capture (shell + workspace watcher) to the framed event stream."""
+
+    def __init__(self, config: AgentConfig) -> None:
+        self.config = config
+        self._spool = Spool(config.spool_path)
+        self._pending: deque[str] = deque()
+        self._seq = 0
+        self._emit_lock = threading.Lock()  # serializes seq + spool + flush
+        self._conn_lock = threading.Lock()  # guards _conn swaps
+        self._conn: WsConnection | None = None
+        self._last_sent: str | None = None  # one-line replay margin, see below
+        self._stop = threading.Event()
+        self._threads: list[threading.Thread] = []
+        self._baseline: dict[str, tuple[int, int, str | None]] = {}
+        self._resume_from_spool()
+
+    # -- durability ------------------------------------------------------
+    def _resume_from_spool(self) -> None:
+        highest = -1
+        for line in self._spool.read_all():
+            self._pending.append(line)
+            try:
+                seq = int(json.loads(line).get("seq", -1))
+            except (ValueError, AttributeError):
+                continue
+            highest = max(highest, seq)
+        self._seq = highest + 1
+
+    # -- event construction ---------------------------------------------
+    def _wire_frame(self, spooled_line: str) -> str:
+        """Spool format -> wire format: strip URL-owned identity fields.
+
+        The spool keeps full events (local durability + restart recovery).
+        The ingest endpoint binds identity at the WS handshake (query params)
+        and rejects frames carrying learner_id/task_id (`extra="forbid"`
+        anti-spoofing), so the wire frame carries only seq/kind/payload/ts.
+        """
+        import json as _json
+
+        full = _json.loads(spooled_line)
+        wire = {
+            k: full[k]
+            for k in ("seq", "kind", "payload", "ts")
+        }
+        if full.get("sandbox_id"):
+            wire["sandbox_id"] = full["sandbox_id"]
+        return _json.dumps(wire)
+
+    def _next_event(self, kind: str, payload: dict[str, Any]) -> dict[str, Any]:
+        if kind not in _EVENT_KINDS:
+            raise ValueError(f"unknown event kind: {kind!r}")
+        event = {
+            "learner_id": self.config.learner_id,
+            "task_id": self.config.task_id,
+            "seq": self._seq,
+            "kind": kind,
+            "payload": payload,
+            "ts": datetime.now(UTC).isoformat(),
+            "sandbox_id": self.config.sandbox_id,
+        }
+        self._seq += 1
+        return event
+
+    # -- emission / flush (D-026) ----------------------------------------
+    def emit(self, kind: str, payload: dict[str, Any]) -> dict[str, Any]:
+        """Spool-then-send. Never blocks on reconnect; loss is impossible."""
+        with self._emit_lock:
+            line = json.dumps(self._next_event(kind, payload))
+            self._spool.append(line)  # durable BEFORE any send attempt
+            self._pending.append(line)
+            self._flush_locked()
+            return json.loads(line)
+
+    def _flush_locked(self) -> None:
+        conn = self._current_conn()
+        while self._pending and conn is not None:
+            line = self._pending[0]
+            try:
+                conn.send_text(self._wire_frame(line))
+            except (ConnectionError, OSError):
+                self._drop_conn()
+                return
+            self._pending.popleft()
+            self._last_sent = line  # kept until a later send proves delivery
+        if not self._pending and self._last_sent is not None:
+            # Compact, but retain the most recently sent line: a send into a
+            # silently-dead socket "succeeds" once at TCP level, so the last
+            # line is only confirmed-sent once a later write works. Retention
+            # is cheap; the server dedups on (learner, task, seq).
+            self._spool.rewrite([self._last_sent])
+
+    def replay_margin(self) -> None:
+        """Requeue the last-sent line after a detected disconnect."""
+        with self._emit_lock:
+            if self._last_sent is not None and (
+                not self._pending or self._pending[0] != self._last_sent
+            ):
+                self._pending.appendleft(self._last_sent)
+                self._spool.rewrite(list(self._pending))
+            self._last_sent = None
+
+    # -- connection supervision ------------------------------------------
+    def _current_conn(self) -> WsConnection | None:
+        with self._conn_lock:
+            return self._conn
+
+    def _set_conn(self, conn: WsConnection | None) -> None:
+        with self._conn_lock:
+            self._conn = conn
+
+    def _drop_conn(self) -> None:
+        conn = self._current_conn()
+        self._set_conn(None)
+        if conn is not None:
+            conn.close()
+        self.replay_margin()
+
+    def is_connected(self) -> bool:
+        return self._current_conn() is not None
+
+    def wait_connected(self, timeout_s: float) -> bool:
+        return self._wait_for(lambda: self.is_connected(), timeout_s)
+
+    def wait_disconnected(self, timeout_s: float) -> bool:
+        return self._wait_for(lambda: not self.is_connected(), timeout_s)
+
+    def _wait_for(self, pred: Any, timeout_s: float) -> bool:
+        deadline = time.monotonic() + timeout_s
+        while time.monotonic() < deadline:
+            if pred():
+                return True
+            time.sleep(0.02)
+        return pred()
+
+    def _supervisor_loop(self) -> None:
+        """Maintain the WS connection: connect, flush backlog, read, backoff."""
+        backoff = self.config.backoff_base_s
+        while not self._stop.is_set():
+            if self._current_conn() is None:
+                try:
+                    conn = WsConnection.connect(self.config.ingest_url)
+                except (ConnectionError, OSError, ValueError, TimeoutError):
+                    self._stop.wait(backoff)
+                    backoff = min(self.config.backoff_max_s, backoff * 2)
+                    continue
+                self._set_conn(conn)
+                self._last_sent = None
+                backoff = self.config.backoff_base_s
+                with self._emit_lock:  # ordered against concurrent emit()s
+                    self._flush_locked()
+            else:
+                conn = self._current_conn()
+            if conn is None:
+                continue
+            try:
+                frame = conn.recv_message(timeout_s=1.0)
+            except (ConnectionError, OSError):
+                self._drop_conn()
+                continue
+            if frame is None:
+                continue
+            opcode, _payload = frame
+            if opcode == 0x8:  # server close frame
+                self._drop_conn()
+
+    # -- workspace watcher ------------------------------------------------
+    def _snapshot_workspace(self) -> dict[str, tuple[int, int, str | None]]:
+        """Map rel path -> (mtime_ns, size, text-or-None-if-too-large)."""
+        snap: dict[str, tuple[int, int, str | None]] = {}
+        root = self.config.workspace
+        if not root.is_dir():
+            return snap
+        for dirpath, dirnames, filenames in os.walk(root):
+            dirnames[:] = sorted(
+                d for d in dirnames if not d.startswith(_AGENT_DIR_PREFIX)
+            )
+            for name in sorted(filenames):
+                if name.startswith(_AGENT_DIR_PREFIX):
+                    continue
+                path = Path(dirpath) / name
+                try:
+                    st = path.stat()
+                except OSError:
+                    continue
+                rel = path.relative_to(root).as_posix()
+                text: str | None = None
+                if st.st_size <= _MAX_DIFF_BYTES:
+                    try:
+                        text = path.read_text(encoding="utf-8", errors="replace")
+                    except OSError:
+                        pass
+                snap[rel] = (st.st_mtime_ns, st.st_size, text)
+        return snap
+
+    def _file_diff_payload(self, rel: str, change: str, old: str | None, new: str | None) -> dict:
+        payload: dict[str, Any] = {"path": rel, "change": change}
+        if old is None and new is None:
+            payload["truncated"] = True
+            return payload
+        diff = "".join(
+            difflib.unified_diff(
+                (old or "").splitlines(keepends=True),
+                (new or "").splitlines(keepends=True),
+                fromfile=f"a/{rel}",
+                tofile=f"b/{rel}",
+            )
+        )
+        payload["diff"] = diff
+        payload["size"] = len(new or "")
+        return payload
+
+    def _watcher_loop(self) -> None:
+        # Baseline is taken in start() before it returns, so any change made
+        # after start() completes is guaranteed to be observed.
+        baseline = self._baseline
+        last_heartbeat = time.monotonic()
+        while not self._stop.wait(self.config.poll_interval_s):
+            current = self._snapshot_workspace()
+            for rel in sorted(current.keys() | baseline.keys()):
+                if rel not in baseline and rel in current:
+                    self.emit(
+                        "file_diff",
+                        self._file_diff_payload(rel, "created", None, current[rel][2]),
+                    )
+                elif rel in baseline and rel not in current:
+                    self.emit(
+                        "file_diff",
+                        self._file_diff_payload(rel, "deleted", baseline[rel][2], None),
+                    )
+                else:
+                    old_stat, new_stat = baseline[rel], current[rel]
+                    if old_stat[:2] != new_stat[:2] and old_stat[2] != new_stat[2]:
+                        self.emit(
+                            "file_diff",
+                            self._file_diff_payload(
+                                rel, "modified", old_stat[2], new_stat[2]
+                            ),
+                        )
+            baseline = current
+            if time.monotonic() - last_heartbeat >= self.config.activity_interval_s:
+                self.emit("activity", {"state": "idle", "spooled": len(self._pending)})
+                last_heartbeat = time.monotonic()
+
+    # -- shell wrapper ------------------------------------------------------
+    @staticmethod
+    def _is_test_command(cmd: str) -> bool:
+        return "test" in cmd.lower()
+
+    def run_command(self, line: str) -> dict[str, Any] | None:
+        """Run one REPL line; emits stdin/command/stdout/run|test_result."""
+        line = line.strip()
+        if not line:
+            return None
+        self.emit("stdin", {"line": line})
+        self.emit("activity", {"state": "command", "spooled": len(self._pending)})
+        self.emit("command", {"cmd": line})
+        started = time.monotonic()
+        timed_out = False
+        exit_code: int | None = None
+        out: str | bytes = ""
+        err: str | bytes = ""
+        try:
+            proc = subprocess.run(
+                ["sh", "-c", line],
+                cwd=self.config.workspace,
+                capture_output=True,
+                timeout=self.config.command_timeout_s,
+                text=True,
+                errors="replace",
+            )
+            exit_code, out, err = proc.returncode, proc.stdout, proc.stderr
+        except subprocess.TimeoutExpired as exc:
+            timed_out = True
+            # TimeoutExpired output attrs are always bytes (even in text mode).
+            out = exc.stdout or b""
+            err = exc.stderr or b""
+        duration = time.monotonic() - started
+        for stream, data in (("stdout", out), ("stderr", err)):
+            if isinstance(data, bytes):
+                data = data.decode(errors="replace")
+            if data:
+                self.emit("stdout", {"stream": stream, "data": data[-_MAX_OUTPUT_CHARS:]})
+        kind = "test_result" if self._is_test_command(line) else "run_result"
+        result = self.emit(
+            kind,
+            {
+                "cmd": line,
+                "exit_code": exit_code,
+                "duration_s": round(duration, 6),
+                "timed_out": timed_out,
+            },
+        )
+        self.emit("activity", {"state": "idle", "spooled": len(self._pending)})
+        return result
+
+    # -- lifecycle ----------------------------------------------------------
+    def start(self) -> None:
+        self.config.workspace.mkdir(parents=True, exist_ok=True)
+        self._baseline = self._snapshot_workspace()
+        self.emit("activity", {"state": "starting", "spooled": len(self._pending)})
+        self._threads = [
+            threading.Thread(target=self._supervisor_loop, daemon=True, name="nc-ws"),
+            threading.Thread(target=self._watcher_loop, daemon=True, name="nc-watch"),
+        ]
+        for thread in self._threads:
+            thread.start()
+
+    def stop(self) -> None:
+        if self._stop.is_set():
+            return
+        try:
+            self.emit("activity", {"state": "stopped", "spooled": len(self._pending)})
+        finally:
+            self._stop.set()
+            self._drop_conn()
+            for thread in self._threads:
+                thread.join(timeout=3)
+            for thread in self._threads:
+                thread.join(timeout=3)
+        with self._emit_lock:
+            self._spool.rewrite(list(self._pending))
+
+
+def main() -> int:
+    try:
+        config = AgentConfig.from_env()
+    except ValueError as exc:
+        print(f"sandbox-agent: {exc}", file=sys.stderr)
+        return 2
+    agent = Agent(config)
+    agent.start()
+    try:
+        # REPL mode: each stdin line is executed and reported (interactive use).
+        # Daemon mode: when stdin is closed/absent (the sandbox backend spawns
+        # the agent with stdin=DEVNULL), keep streaming workspace diffs +
+        # activity until SIGTERM/SIGINT so the agent's lifecycle is tied to
+        # the sandbox (destroy() reaps it) rather than to stdin EOF.
+        if sys.stdin is None or sys.stdin.closed:  # pragma: no cover - defensive
+            agent._stop.wait()  # noqa: SLF001 - daemon block
+        else:
+            line = sys.stdin.readline()
+            while line:
+                agent.run_command(line)
+                line = sys.stdin.readline()
+            if not agent._stop.is_set() and not sys.stdin.isatty():  # noqa: SLF001
+                # EOF on a pipe (DEVNULL): daemonize — watch + stream until killed.
+                import signal
+
+                signal.signal(signal.SIGTERM, lambda *_: agent._stop.set())  # noqa: SLF001
+                agent._stop.wait()  # noqa: SLF001
+    except KeyboardInterrupt:
+        pass
+    finally:
+        agent.stop()
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

apps/ai-service/tests/agents/test_assessor.py

@@ -1,103 +1,118 @@
-"""Assessor agent tests — structured rubric scores (REQ-2-008).
+"""Assessor agent tests — live-grade coaching contract (REQ-3-007).
 
-The Assessor is the structured-output showcase: tests use ScriptedJSONProvider
-for valid payloads and exercise the 4-layer defense failure modes.
+v0.3 re-grounding: the Assessor renders coaching FROM the stored grade
+(GradeRecord) — it never invents scores (the grading engine owns that).
 """
 
+from __future__ import annotations
+
+import json
+from datetime import UTC, datetime
+
 import pytest
 
-from ai_service.agents.assessor import AssessorAgent, RubricScore
-from ai_service.agents.structured import StructuredOutputError
+from ai_service.agents.assessor import AssessorAgent, GradeCoaching
 from ai_service.config import Settings
-from ai_service.corpus.artifacts import (
-    get_artifact_bundle,
-    get_transcript_for_artifact,
-    render_rubric,
+from ai_service.grading.store import GradeRecord, SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.llm.types import Message
+
+COACHING_JSON = json.dumps(
+    {
+        "summary": "Solid iterative build; tests drove the fixes.",
+        "strengths": ["Ran tests after each change."],
+        "gaps": ["Did not cover the empty-input case."],
+        "next_steps": ["Add one edge-case test."],
+    }
 )
-from ai_service.corpus.learner_context import get_learner_context
-from ai_service.llm.mock import MockProvider, ScriptedJSONProvider
-
-VALID_SCORE = {
-    "rubric_id": "rubric-orchestration-c002",
-    "artifact_id": "art-eval-research-assistant",
-    "competency_id": "stack-orchestration-c002",
-    "scores": [
-        {"criterion_id": "rc-architecture", "name": "Agent architecture soundness",
-         "score": 92, "evidence": "Explicit state schema with planner-only write access"},
-        {"criterion_id": "rc-communication", "name": "Inter-agent communication design",
-         "score": 88, "evidence": "Typed ToolMessage responses with retry flags"},
-        {"criterion_id": "rc-reliability", "name": "Reliability engineering",
-         "score": 85, "evidence": "3-retry loop with degradation path"},
-        {"criterion_id": "rc-process", "name": "Process trace quality",
-         "score": 90, "evidence": "Iterative saves with passing test checkpoints"},
-    ],
-    "strengths": ["Clean state boundaries", "Failure-aware tool wrapping"],
-    "gaps": ["No reviewer node yet", "Graph diagram only in README"],
-    "verdict": "mastered",
-}
 
 
-def make_assessor(provider=None) -> AssessorAgent:
-    return AssessorAgent(provider or MockProvider(), Settings(provider="mock"))
+class ScriptedProvider(MockProvider):
+    def __init__(self) -> None:
+        super().__init__()
+        self.requests: list[list[Message]] = []
+        self.replies: list[str] = []
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):
+        self.requests.append([Message(role=m.role, content=m.content) for m in messages])
+        if self.replies:
+            return self.replies.pop(0)
+        return COACHING_JSON
 
 
-def get_bundle(artifact_id="art-eval-research-assistant"):
-    bundle = get_artifact_bundle(artifact_id)
-    assert bundle is not None
-    return bundle
-
-
-async def test_evaluate_returns_validated_rubric_score():
-    provider = ScriptedJSONProvider(VALID_SCORE)
-    assessor = make_assessor(provider)
-    artifact, rubric = get_bundle()
-    transcript = get_transcript_for_artifact(artifact.artifact_id)
-    result = await assessor.evaluate(artifact, rubric, transcript)
-    assert isinstance(result, RubricScore)
-    assert result.verdict == "mastered"
-    assert len(result.scores) == 4
-    assert result.weighted_total(rubric) == pytest.approx(
-        92 * 0.3 + 88 * 0.3 + 85 * 0.25 + 90 * 0.15
+def _grade() -> GradeRecord:
+    return GradeRecord(
+        learner_id="assessor-learner",
+        task_id="assessor-task",
+        variant_seed=None,
+        digest={"error_fix_cycles": 2, "final_test_status": "pass"},
+        scores={
+            "criteria": {
+                "process_quality": 4,
+                "correctness": 3,
+                "debugging_discipline": 4,
+                "test_usage": 3,
+            },
+            "strengths": ["s"],
+            "gaps": ["g"],
+            "verdict": "developing",
+        },
+        verdict="GRADED",
+        model="gemma4:31b",
+        created_at=datetime.now(UTC),
     )
 
 
-async def test_evaluate_rejects_invalid_schema_after_retry():
-    """Plain MockProvider returns non-rubric JSON → 4-layer defense exhausts
-    its single retry and raises StructuredOutputError."""
-    assessor = make_assessor(MockProvider())
-    artifact, rubric = get_bundle()
-    transcript = get_transcript_for_artifact(artifact.artifact_id)
-    with pytest.raises(StructuredOutputError):
-        await assessor.evaluate(artifact, rubric, transcript)
+@pytest.fixture()
+def provider() -> ScriptedProvider:
+    return ScriptedProvider()
 
 
-def test_build_evaluation_input_carries_all_inputs():
-    assessor = make_assessor()
-    artifact, rubric = get_bundle()
-    transcript = get_transcript_for_artifact(artifact.artifact_id)
-    text = assessor.build_evaluation_input(artifact, rubric, transcript)
-    assert artifact.name in text
-    assert artifact.evidence_excerpt in text
-    assert "rc-architecture" in text  # rubric rendered
-    assert "examiner:" in text  # transcript rendered
+@pytest.fixture()
+def agent(provider) -> AssessorAgent:
+    return AssessorAgent(provider, Settings(provider="mock"))
 
 
-def test_build_evaluation_input_without_transcript():
-    assessor = make_assessor()
-    artifact, rubric = get_bundle()
-    text = assessor.build_evaluation_input(artifact, rubric, None)
-    assert artifact.name in text
-    assert "examiner:" not in text
+class TestCoachGrade:
+    async def test_prompt_contains_stored_grade_not_learner_id(self, agent, provider) -> None:
+        await agent.coach_grade(_grade())
+        all_text = "\n".join(
+            m.content for request in provider.requests for m in request
+        )
+        assert "process_quality" in all_text  # stored scores rendered
+        assert "GRADED" in all_text
+        assert "assessor-learner" not in all_text  # D-028 anonymity
+
+    async def test_coaching_validates_via_d020(self, agent, provider) -> None:
+        coaching = await agent.coach_grade(_grade())
+        assert isinstance(coaching, GradeCoaching)
+        assert coaching.summary
+        assert coaching.next_steps
+
+    async def test_malformed_then_good_exercises_retry(self, agent, provider) -> None:
+        provider.replies = ["garbage", COACHING_JSON]
+        coaching = await agent.coach_grade(_grade())
+        assert coaching.summary
+        assert len(provider.requests) == 2
+
+    async def test_no_corpus_artifact_imports(self) -> None:
+        import ast
+        from pathlib import Path
+
+        py = Path(__file__).parents[2] / "ai_service" / "agents" / "assessor.py"
+        tree = ast.parse(py.read_text())
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ImportFrom) and node.module:
+                assert "corpus.artifacts" not in node.module
+                assert "corpus.telemetry" not in node.module
 
 
-def test_system_prompt_names_assessor_persona():
-    prompt = make_assessor().system_prompt(get_learner_context())
-    assert "Assessor" in prompt
-    assert "ONLY" in prompt  # JSON-only instruction
-
-
-def test_rubric_render_in_prompt_is_complete():
-    """The rubric passed to the model lists every criterion (fair grading)."""
-    artifact, rubric = get_bundle()
-    text = render_rubric(rubric)
-    assert text.count("rc-") == len(rubric.criteria)
+class TestStoreRoundtrip:
+    def test_grade_store_roundtrip(self, tmp_path) -> None:
+        store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+        record = _grade()
+        store.save(record)
+        fetched = store.get("assessor-learner", "assessor-task")
+        assert fetched is not None
+        assert fetched.scores["criteria"]["process_quality"] == 4
+        store.close()

apps/ai-service/tests/agents/test_examiner.py

@@ -0,0 +1,159 @@
+"""Examiner agent tests (Task 5-2-01, REQ-3-006)."""
+
+from __future__ import annotations
+
+import json
+
+import pytest
+
+from ai_service.agents.examiner import DefenseVerdict, ExaminerAgent
+from ai_service.agents.registry import AgentRegistry, register_builtin_agents
+from ai_service.config import Settings
+from ai_service.grading.features import compute_digest
+from ai_service.llm.mock import MockProvider
+from ai_service.llm.types import Message
+
+VERDICT_JSON = json.dumps(
+    {
+        "verdict": "developing",
+        "understanding": "Explains the retry loop clearly.",
+        "process_justification": "Justifies the edit-then-test cadence from the digest.",
+        "communication": "Answers are specific and on-topic.",
+        "strengths": ["Grounded the fix in a failed test."],
+        "gaps": ["Did not justify the chunk-size choice."],
+    }
+)
+
+
+class RecordingProvider(MockProvider):
+    """Mock provider that records every message list (prompt assertions)."""
+
+    def __init__(self, replies: list[str] | None = None) -> None:
+        super().__init__()
+        self.replies = list(replies or [])
+        self.requests: list[list[Message]] = []
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):
+        self.requests.append([Message(role=m.role, content=m.content) for m in messages])
+        if self.replies:
+            return self.replies.pop(0)
+        return "Tell me about your build."
+
+
+def _digest():
+    from datetime import UTC, datetime, timedelta
+
+    from ai_service.telemetry.models import TelemetryEvent
+
+    t0 = datetime(2026, 9, 12, tzinfo=UTC)
+    events = [
+        TelemetryEvent(
+            learner_id="examiner-learner",
+            task_id="examiner-task",
+            seq=n,
+            kind=kind,
+            payload=payload,
+            ts=t0 + timedelta(seconds=n * 10),
+            sandbox_id="sbx-examiner",
+        )
+        for n, (kind, payload) in enumerate(
+            [
+                ("file_diff", {"path": "a.py"}),
+                ("command", {"cmd": "pytest -q"}),
+                ("test_result", {"passed": False, "exit_code": 1}),
+                ("file_diff", {"path": "a.py"}),
+                ("test_result", {"passed": True, "exit_code": 0}),
+            ]
+        )
+    ]
+    return compute_digest(events)
+
+
+@pytest.fixture()
+def provider() -> RecordingProvider:
+    return RecordingProvider()
+
+
+@pytest.fixture()
+def settings() -> Settings:
+    return Settings(provider="mock")
+
+
+@pytest.fixture()
+def examiner(provider, settings) -> ExaminerAgent:
+    return ExaminerAgent(provider, settings)
+
+
+class TestNextQuestion:
+    async def test_prompt_contains_digest_but_no_learner_id(
+        self, examiner, provider
+    ) -> None:
+        await examiner.next_question(
+            history=[Message(role="assistant", content="First question?")],
+            trace_digest=_digest(),
+            variant_statement="Build a chunker.",
+        )
+        all_content = "\n".join(
+            m.content for request in provider.requests for m in request
+        )
+        assert "error_fix_cycles" in all_content  # digest JSON grounded
+        assert "examiner-learner" not in all_content  # D-028 anonymity
+        assert "Build a chunker." in all_content  # variant statement grounded
+        assert all_content.count('"examiner-learner"') == 0
+
+    async def test_question_returned_from_provider(self, examiner) -> None:
+        question = await examiner.next_question(
+            history=[], trace_digest=_digest()
+        )
+        assert isinstance(question, str)
+
+
+class TestFinalVerdict:
+    async def test_verdict_validates_via_d020(self, examiner, provider) -> None:
+        provider.replies = [VERDICT_JSON]
+        verdict = await examiner.final_verdict(
+            history=[Message(role="assistant", content="Q?")],
+            trace_digest=_digest(),
+        )
+        assert isinstance(verdict, DefenseVerdict)
+        assert verdict.verdict == "developing"
+        assert verdict.strengths and verdict.gaps
+
+    async def test_malformed_then_good_exercises_retry(self, examiner, provider) -> None:
+        provider.replies = ["not json", VERDICT_JSON]
+        verdict = await examiner.final_verdict(history=[], trace_digest=_digest())
+        assert verdict.verdict == "developing"
+        assert len(provider.requests) == 2  # D-020 bounded retry
+
+
+class TestRegistry:
+    def test_all_seven_agents_resolve(self, provider, settings) -> None:
+        registry = AgentRegistry()
+        register_builtin_agents(registry)
+        assert registry.names() == [
+            "assessor",
+            "coach",
+            "examiner",
+            "lab",
+            "mentor",
+            "proctor",
+            "tutor",
+        ]
+        agent = registry.get(provider, settings, "examiner")
+        assert isinstance(agent, ExaminerAgent)
+
+
+class TestBoundary:
+    def test_examiner_never_imports_voice_or_api(self) -> None:
+        import ast
+        from pathlib import Path
+
+        py = Path(__file__).parents[2] / "ai_service" / "agents" / "examiner.py"
+        tree = ast.parse(py.read_text())
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ImportFrom) and node.module:
+                assert "voice" not in node.module, "examiner must not import voice/"
+                assert not node.module.startswith("ai_service.api")
+            if isinstance(node, ast.Import):
+                for alias in node.names:
+                    assert alias.name != "fastapi"

apps/ai-service/tests/agents/test_registry.py

@@ -97,10 +97,11 @@ def test_proctor_and_mentor_resolve_via_registry():
     assert isinstance(mentor, MentorAgent)
 
 
-def test_registry_resolves_all_six_agents():
-    """Must-Have (Phase 5): the full roster — coach/tutor/lab/assessor/proctor/mentor."""
+def test_registry_resolves_all_seven_agents():
+    """Must-Have (Phase 5): the full roster — six tutors + the Examiner."""
     from ai_service.agents.assessor import AssessorAgent
     from ai_service.agents.coach import CoachAgent
+    from ai_service.agents.examiner import ExaminerAgent
     from ai_service.agents.lab import LabAgent
     from ai_service.agents.mentor import MentorAgent
     from ai_service.agents.proctor import ProctorAgent
@@ -108,7 +109,9 @@ def test_registry_resolves_all_six_agents():
 
     registry = AgentRegistry()
     register_builtin_agents(registry)
-    assert registry.names() == ["assessor", "coach", "lab", "mentor", "proctor", "tutor"]
+    assert registry.names() == [
+        "assessor", "coach", "examiner", "lab", "mentor", "proctor", "tutor",
+    ]
     settings = Settings(provider="mock")
     expected = {
         "coach": CoachAgent,
@@ -117,6 +120,7 @@ def test_registry_resolves_all_six_agents():
         "assessor": AssessorAgent,
         "proctor": ProctorAgent,
         "mentor": MentorAgent,
+        "examiner": ExaminerAgent,
     }
     for name, cls in expected.items():
         agent = registry.get(MockProvider(), settings, name)

apps/ai-service/tests/agents/test_regrounding_live.py

@@ -0,0 +1,207 @@
+"""Live re-grounding tests: Lab, Assessor, Proctor on REAL inputs (REQ-3-007)."""
+
+from __future__ import annotations
+
+import json
+import tempfile
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+from fastapi.testclient import TestClient
+
+from ai_service.agents.assessor import AssessorAgent, GradeCoaching
+from ai_service.agents.lab import LabAgent
+from ai_service.agents.proctor import ProctorAgent, ProctorAssessment
+from ai_service.config import Settings
+from ai_service.grading.store import GradeRecord, SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.llm.types import Message
+from ai_service.telemetry.models import TelemetryEvent
+from ai_service.telemetry.store import SQLiteTraceStore
+
+COACHING_JSON = json.dumps(
+    {
+        "summary": "Iterative build with test discipline.",
+        "strengths": ["Tested after changes."],
+        "gaps": ["Missing edge cases."],
+        "next_steps": ["Add an edge-case test."],
+    }
+)
+PROCTOR_JSON = json.dumps(
+    {
+        "signals": [
+            {"signal_type": "idle_gap", "severity": "low", "note": "One 400s gap."}
+        ],
+        "intervention": "Offer a short break.",
+        "summary": "Healthy session overall.",
+    }
+)
+T0 = datetime(2026, 9, 12, tzinfo=UTC)
+
+
+class RecordingProvider(MockProvider):
+    def __init__(self, structured_json: str) -> None:
+        super().__init__()
+        self._structured_json = structured_json
+        self.requests: list[list[Message]] = []
+
+    def _reply_for(self, messages, response_format):
+        self.requests.append([Message(role=m.role, content=m.content) for m in messages])
+        if response_format is not None and response_format.get("type") == "json_object":
+            return self._structured_json
+        return "Coaching feedback referencing your latest test run."
+
+
+def _event(
+    seq: int, kind: str, payload: dict, offset_s: float,
+    learner="live-learner", task="live-task",
+):
+    return TelemetryEvent(
+        learner_id=learner,
+        task_id=task,
+        seq=seq,
+        kind=kind,
+        payload=payload,
+        ts=T0 + timedelta(seconds=offset_s),
+        sandbox_id="sbx-live",
+    )
+
+
+def _seed_trace(store: SQLiteTraceStore) -> None:
+    events = [
+        _event(0, "file_diff", {"path": "a.py"}, 0),
+        _event(1, "command", {"cmd": "pytest -q"}, 10),
+        _event(2, "test_result", {"passed": False, "exit_code": 1}, 15),
+        _event(3, "file_diff", {"path": "a.py"}, 30),
+        _event(4, "test_result", {"passed": True, "exit_code": 0}, 45),
+        _event(5, "activity", {"state": "idle"}, 500),  # >120s gap -> idle
+    ]
+    for e in events:
+        store.append(e)
+
+
+class TestLabLive:
+    async def test_lab_prompt_contains_digest_not_corpus(self, tmp_path) -> None:
+        store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+        _seed_trace(store)
+        provider = RecordingProvider("feedback")
+        agent = LabAgent(provider, Settings(provider="mock"))
+
+        from ai_service.grading.features import compute_digest
+
+        digest = compute_digest(store.get_trace("live-learner", "live-task"))
+        tokens = [t async for t in agent.stream_feedback(digest)]
+        assert tokens
+        all_text = "\n".join(
+            m.content for request in provider.requests for m in request
+        )
+        assert "error_fix_cycles" in all_text
+        assert "lab-scenario" not in all_text  # no corpus fixture ids
+        store.close()
+
+    def test_no_corpus_telemetry_import_in_lab(self) -> None:
+        import ast
+        from pathlib import Path
+
+        py = Path(__file__).parents[2] / "ai_service" / "agents" / "lab.py"
+        tree = ast.parse(py.read_text())
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ImportFrom) and node.module:
+                assert "corpus.telemetry" not in node.module
+
+
+class TestAssessorLive:
+    async def test_assessor_prompt_contains_stored_scores(self) -> None:
+        provider = RecordingProvider(COACHING_JSON)
+        agent = AssessorAgent(provider, Settings(provider="mock"))
+        grade = GradeRecord(
+            learner_id="live-learner",
+            task_id="live-task",
+            variant_seed=None,
+            digest={"error_fix_cycles": 1},
+            scores={"criteria": {"process_quality": 3}},
+            verdict="GRADED",
+            model="mock",
+            created_at=datetime.now(UTC),
+        )
+        coaching = await agent.coach_grade(grade)
+        assert isinstance(coaching, GradeCoaching)
+        all_text = "\n".join(
+            m.content for request in provider.requests for m in request
+        )
+        assert "process_quality" in all_text
+        assert "live-learner" not in all_text
+
+    def test_no_corpus_artifact_import_in_assessor(self) -> None:
+        import ast
+        from pathlib import Path
+
+        py = Path(__file__).parents[2] / "ai_service" / "agents" / "assessor.py"
+        tree = ast.parse(py.read_text())
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ImportFrom) and node.module:
+                assert "corpus.artifacts" not in node.module
+                assert "corpus.telemetry" not in node.module
+
+
+class TestProctorLive:
+    async def test_proctor_receives_real_digest_and_defense_signals(self) -> None:
+        provider = RecordingProvider(PROCTOR_JSON)
+        agent = ProctorAgent(provider, Settings(provider="mock"))
+        from ai_service.grading.features import compute_digest
+
+        store = SQLiteTraceStore(db_path=Path(tempfile.mkdtemp()) / "proctor-t.db")
+        _seed_trace(store)
+        digest = compute_digest(store.get_trace("live-learner", "live-task"))
+        store.close()
+
+        assessment = await agent.assess(
+            digest,
+            defense_signals={"long_pauses": [{"turn": 3, "latency_ms": 30000}]},
+            variant_context={"template_id": "tpl-llm-judge", "seed": "cafe", "params": {}},
+        )
+        assert isinstance(assessment, ProctorAssessment)
+        all_text = "\n".join(
+            m.content for request in provider.requests for m in request
+        )
+        assert "idle_gap" in all_text or "idle_gap_count" in all_text
+        assert "long_pauses" in all_text
+        assert "tpl-llm-judge" in all_text
+        assert "live-learner" not in all_text
+
+    def test_no_corpus_scenario_import_in_proctor(self) -> None:
+        import ast
+        from pathlib import Path
+
+        py = Path(__file__).parents[2] / "ai_service" / "agents" / "proctor.py"
+        tree = ast.parse(py.read_text())
+        for node in ast.walk(tree):
+            if isinstance(node, ast.ImportFrom) and node.module:
+                assert "corpus.telemetry" not in node.module
+
+
+class TestProctorEndpoint:
+    def test_signals_endpoint_serves_real_inputs(self, tmp_path) -> None:
+        from ai_service.main import create_app
+        from ai_service.telemetry.ingest import TraceIntegrityMap
+        from ai_service.variants.store import SQLiteVariantStore
+        from ai_service.voice.defense_store import SQLiteDefenseStore
+
+        app = create_app(Settings(provider="mock"))
+        provider = RecordingProvider(PROCTOR_JSON)
+        app.state.provider = provider
+        app.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+        app.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+        app.state.variant_store = SQLiteVariantStore(db_path=tmp_path / "v.db")
+        app.state.defense_store = SQLiteDefenseStore(db_path=tmp_path / "d.db")
+        app.state.trace_integrity = TraceIntegrityMap()
+        _seed_trace(app.state.trace_store)
+        with TestClient(app) as client:
+            resp = client.post(
+                "/v1/proctor/signals",
+                json={"learner_id": "live-learner", "task_id": "live-task"},
+            )
+            assert resp.status_code == 200, resp.text
+            body = resp.json()
+            assert body["signals"] is not None
+            assert "intervention" in body

apps/ai-service/tests/api/test_assessment.py

@@ -1,79 +1,87 @@
-"""Assessment evaluate endpoint tests — validated JSON, 404s (REQ-2-008)."""
+"""Assessment API tests — stored-grade coaching contract (REQ-3-007)."""
+
+from __future__ import annotations
+
+import json
+from datetime import UTC, datetime
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.config import Settings
+from ai_service.grading.store import GradeRecord, SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+
+COACHING_JSON = json.dumps(
+    {
+        "summary": "Good iterative work.",
+        "strengths": ["Tests after changes."],
+        "gaps": ["Missing edge cases."],
+        "next_steps": ["Add an edge-case test."],
+    }
+)
 
 
-from ai_service.agents.assessor import RubricScore
-from ai_service.llm.mock import ScriptedJSONProvider
-
-VALID_SCORE = {
-    "rubric_id": "rubric-orchestration-c002",
-    "artifact_id": "art-eval-research-assistant",
-    "competency_id": "stack-orchestration-c002",
-    "scores": [
-        {"criterion_id": "rc-architecture", "name": "Agent architecture soundness",
-         "score": 92, "evidence": "Explicit state schema"},
-        {"criterion_id": "rc-communication", "name": "Inter-agent communication design",
-         "score": 88, "evidence": "Typed ToolMessage responses"},
-        {"criterion_id": "rc-reliability", "name": "Reliability engineering",
-         "score": 85, "evidence": "3-retry loop"},
-        {"criterion_id": "rc-process", "name": "Process trace quality",
-         "score": 90, "evidence": "Iterative checkpoints"},
-    ],
-    "strengths": ["Clean state boundaries", "Failure-aware tools"],
-    "gaps": ["No reviewer node", "Diagram only in README"],
-    "verdict": "mastered",
-}
+class CoachingMock(MockProvider):
+    def _reply_for(self, messages, response_format):
+        if response_format is not None and response_format.get("type") == "json_object":
+            return COACHING_JSON
+        return super()._reply_for(messages, response_format)
 
 
-def test_evaluate_returns_validated_rubric_json(client):
-    # Swap the app provider for a scripted-JSON provider for this test
-    original = client.app.state.provider
-    client.app.state.provider = ScriptedJSONProvider(VALID_SCORE)
-    try:
-        response = client.post(
-            "/v1/assessment/evaluate",
-            json={"artifact_id": "art-eval-research-assistant"},
+@pytest.fixture()
+def client(tmp_path) -> TestClient:
+    app = create_app(Settings(provider="mock"))
+    app.state.provider = CoachingMock()
+    app.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+    app.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+    app.state.trace_integrity = TraceIntegrityMap()
+    with TestClient(app) as c:
+        yield c
+
+
+def _seed_grade(client: TestClient) -> None:
+
+    app = client.app
+    store: SQLiteGradeStore = app.state.grade_store
+    store.save(
+        GradeRecord(
+            learner_id="api-learner",
+            task_id="api-task",
+            variant_seed=None,
+            digest={"error_fix_cycles": 1},
+            scores={"criteria": {"process_quality": 3}, "verdict": "developing"},
+            verdict="GRADED",
+            model="gemma4:31b",
+            created_at=datetime.now(UTC),
         )
-    finally:
-        client.app.state.provider = original
-    assert response.status_code == 200
-    data = response.json()
-    validated = RubricScore.model_validate(data)  # response contract holds
-    assert validated.verdict == "mastered"
-    assert len(validated.scores) == 4
-
-
-def test_unknown_artifact_404(client):
-    response = client.post("/v1/assessment/evaluate", json={"artifact_id": "ghost"})
-    assert response.status_code == 404
-    assert "ghost" in response.json()["detail"]
-
-
-def test_unparseable_provider_502(client):
-    """Plain MockProvider yields non-rubric JSON → structured defense exhausts
-    retry → endpoint translates to 502 (bad gateway to the model)."""
-    # default mock already returns non-rubric JSON
-    response = client.post(
-        "/v1/assessment/evaluate",
-        json={"artifact_id": "art-eval-research-assistant"},
     )
-    assert response.status_code == 502
-    assert "failed" in response.json()["detail"].lower()
 
 
-def test_missing_artifact_id_422(client):
-    response = client.post("/v1/assessment/evaluate", json={})
-    assert response.status_code == 422
+def test_evaluate_renders_stored_grade_as_coaching(client) -> None:
+    _seed_grade(client)
+    resp = client.post(
+        "/v1/assessment/evaluate",
+        json={"learner_id": "api-learner", "task_id": "api-task"},
+    )
+    assert resp.status_code == 200, resp.text
+    body = resp.json()
+    assert body["grade_verdict"] == "GRADED"
+    assert body["coaching"]["summary"]
 
 
-def test_second_artifact_also_evaluates(client):
-    original = client.app.state.provider
-    payload = dict(VALID_SCORE, artifact_id="art-eval-rag-dashboard")
-    client.app.state.provider = ScriptedJSONProvider(payload)
-    try:
-        response = client.post(
-            "/v1/assessment/evaluate", json={"artifact_id": "art-eval-rag-dashboard"}
-        )
-    finally:
-        client.app.state.provider = original
-    assert response.status_code == 200
-    assert response.json()["artifact_id"] == "art-eval-rag-dashboard"
+def test_evaluate_without_grade_404(client) -> None:
+    resp = client.post(
+        "/v1/assessment/evaluate",
+        json={"learner_id": "nobody", "task_id": "nothing"},
+    )
+    assert resp.status_code == 404
+    assert "grade first" in resp.json()["detail"]
+
+
+def test_missing_fields_422(client) -> None:
+    resp = client.post("/v1/assessment/evaluate", json={"learner_id": "x"})
+    assert resp.status_code == 422

apps/ai-service/tests/api/test_cors.py

@@ -0,0 +1,82 @@
+"""CORS policy tests (A-008, D-038 network mode).
+
+v0.3 initially shipped `allow_methods` WITHOUT "PUT" while the learner
+build surface writes workspace files with PUT (engine-client writeFile) —
+every cross-origin Save failed preflight. These tests pin the policy so a
+future method-list edit fails loudly instead of silently breaking the
+headline flow.
+
+v0.3.5 network mode (D-038): the default AI_CORS_ORIGINS='*' admits any
+origin (safe ONLY because credentials are never enabled); an explicit list
+restricts. Both modes are pinned here:
+  - wildcard: remote origin gets the grant; credentials still never sent;
+  - explicit: unlisted origins get no grant.
+"""
+
+from __future__ import annotations
+
+from fastapi.testclient import TestClient
+
+ALLOWED_ORIGIN = "http://localhost:3000"
+REMOTE_ORIGIN = "http://nextcraft-1:3000"
+ALL_CLIENT_METHODS = ("GET", "POST", "PUT", "DELETE")
+
+
+def _allow_origin(resp) -> str | None:
+    return resp.headers.get("access-control-allow-origin")
+
+
+def test_preflight_allows_every_method_the_web_client_uses(client: TestClient) -> None:
+    for method in ALL_CLIENT_METHODS:
+        resp = client.options(
+            "/v1/sandboxes",
+            headers={
+                "Origin": ALLOWED_ORIGIN,
+                "Access-Control-Request-Method": method,
+            },
+        )
+        assert resp.status_code == 200, f"preflight {method} failed: {resp.status_code}"
+        assert _allow_origin(resp) in ("*", ALLOWED_ORIGIN)
+        allowed = resp.headers["access-control-allow-methods"].split(", ")
+        assert method in allowed, f"{method} missing from CORS methods: {allowed}"
+
+
+def test_cross_origin_get_echoes_allow_origin(client: TestClient) -> None:
+    resp = client.get("/v1/sandboxes", headers={"Origin": ALLOWED_ORIGIN})
+    assert resp.status_code == 200
+    assert _allow_origin(resp) in ("*", ALLOWED_ORIGIN)
+
+
+def test_wildcard_mode_grants_remote_origins(client: TestClient) -> None:
+    """D-038 default: '*' grants any origin — remote browsers work zero-config."""
+    resp = client.get("/v1/sandboxes", headers={"Origin": REMOTE_ORIGIN})
+    assert resp.status_code == 200
+    assert _allow_origin(resp) in ("*", REMOTE_ORIGIN)
+
+
+def test_explicit_list_mode_denies_unlisted_origins(
+    settings, monkeypatch, tmp_path
+) -> None:
+    """Explicit AI_CORS_ORIGINS restricts to the listed origins only."""
+    from fastapi.testclient import TestClient as TC
+
+    from ai_service.main import create_app
+
+    restricted = settings.model_copy(update={"cors_origins": "http://localhost:3000"})
+    app = create_app(restricted)
+    with TC(app) as c:
+        resp = c.get("/v1/sandboxes", headers={"Origin": "https://evil.example"})
+        assert resp.status_code == 200  # non-CORS requests still serve
+        assert resp.headers.get("access-control-allow-origin") is None
+
+
+def test_credentials_never_allowed(client: TestClient) -> None:
+    resp = client.options(
+        "/v1/sandboxes",
+        headers={
+            "Origin": ALLOWED_ORIGIN,
+            "Access-Control-Request-Method": "PUT",
+            "Access-Control-Request-Headers": "Content-Type",
+        },
+    )
+    assert resp.headers.get("access-control-allow-credentials") != "true"

apps/ai-service/tests/api/test_defense.py

@@ -0,0 +1,234 @@
+"""Defense endpoint tests (Task 5-3-01, REQ-3-006) — mock voice + mock LLM."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.agents.examiner import ExaminerAgent
+from ai_service.config import Settings
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+from ai_service.variants.store import SQLiteVariantStore
+from ai_service.voice.defense_store import SQLiteDefenseStore
+from ai_service.voice.mock import MockVoiceProvider
+
+VERDICT = {
+    "verdict": "developing",
+    "understanding": "Explains the build clearly.",
+    "process_justification": "Justifies choices.",
+    "communication": "Clear and specific.",
+    "strengths": ["Grounded answers in the digest."],
+    "gaps": ["Did not address the edge cases."],
+}
+
+
+class ScriptedLLM(MockProvider):
+    """Question-mode calls get a question; verdict-mode calls get D-020 JSON.
+
+    Discriminator: the verdict prompt contains "final verdict JSON" — the
+    question prompt says "next question".
+    """
+
+    def __init__(self) -> None:
+        super().__init__()
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):
+        all_text = "\n".join(m.content for m in messages)
+        if "final verdict JSON" in all_text:
+            return json.dumps(VERDICT)
+        return "Why did you structure the fix that way?"
+
+
+@pytest.fixture()
+def app(tmp_path: Path):
+    application = create_app(Settings(provider="mock", voice_provider="mock"))
+    llm = ScriptedLLM()
+    application.state.provider = llm
+    application.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+    application.state.variant_store = SQLiteVariantStore(db_path=tmp_path / "v.db")
+    application.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+    application.state.trace_integrity = TraceIntegrityMap()
+    application.state.defense_store = SQLiteDefenseStore(db_path=tmp_path / "d.db")
+    application.state.voice_provider = MockVoiceProvider(
+        ["the fix was in the retry loop"]
+    )
+    settings = Settings(provider="mock", voice_provider="mock")
+    application.state.examiner_agent = ExaminerAgent(llm, settings)
+    return application
+
+
+@pytest.fixture()
+def client(app) -> TestClient:
+    with TestClient(app) as c:
+        yield c
+
+
+def _start(client: TestClient) -> dict:
+    resp = client.post(
+        "/v1/defense/start", json={"learner_id": "defense-learner", "task_id": "defense-task"}
+    )
+    assert resp.status_code == 200, resp.text
+    return resp.json()
+
+
+class TestStart:
+    def test_start_returns_first_question_and_descriptor(self, client) -> None:
+        body = _start(client)
+        assert body["first_question"]
+        assert body["defense_id"]
+        assert body["voice_descriptor"]["mode"] == "mock"
+        assert body["trace_complete"] is True
+        stored = client.get(f"/v1/defense/{body['defense_id']}")
+        assert stored.status_code == 200
+        turns = stored.json()["turns"]
+        assert turns and turns[0]["role"] == "examiner"
+
+    def test_start_with_unknown_trace_is_complete_flag(self, client) -> None:
+        body = _start(client)
+        assert body["trace_complete"] is True
+
+
+class TestBrowserFallback:
+    def test_browser_mode_serves_browser_descriptor(self, tmp_path: Path) -> None:
+        """Must-Have #6: AI_VOICE_PROVIDER=browser → start returns the
+        browser-native SR/TTS fallback descriptor (D-030), not 'mock'."""
+        application = create_app(Settings(provider="mock", voice_provider="browser"))
+        application.state.provider = ScriptedLLM()
+        application.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+        application.state.variant_store = SQLiteVariantStore(db_path=tmp_path / "v.db")
+        application.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+        application.state.trace_integrity = TraceIntegrityMap()
+        application.state.defense_store = SQLiteDefenseStore(db_path=tmp_path / "d.db")
+        application.state.voice_provider = MockVoiceProvider(["answer"])
+        llm = ScriptedLLM()
+        application.state.examiner_agent = ExaminerAgent(
+            llm, Settings(provider="mock", voice_provider="browser")
+        )
+        with TestClient(application) as c:
+            body = _start(c)
+            assert body["voice_descriptor"]["mode"] == "browser"
+            assert body["voice_descriptor"]["sr_available"]
+            assert "SpeechRecognition" in body["voice_descriptor"]["hint"]
+
+
+class TestAnswer:
+    def test_typed_answer_yields_followup_with_latency(self, client) -> None:
+        defense_id = _start(client)["defense_id"]
+        resp = client.post(
+            f"/v1/defense/{defense_id}/answer", data={"text": "I fixed the loop."}
+        )
+        assert resp.status_code == 200, resp.text
+        body = resp.json()
+        assert body["question"]
+        assert body["turn_latency"]["llm_ms"] is not None
+
+    def test_audio_answer_transcribed_and_recorded(self, client) -> None:
+        defense_id = _start(client)["defense_id"]
+        wav_bytes = b"RIFF" + b"\x00" * 64
+        resp = client.post(
+            f"/v1/defense/{defense_id}/answer",
+            files={"audio": ("answer.wav", wav_bytes, "audio/wav")},
+        )
+        assert resp.status_code == 200, resp.text
+        stored = client.get(f"/v1/defense/{defense_id}").json()
+        learner_turns = [t for t in stored["turns"] if t["role"] == "learner"]
+        assert learner_turns, "learner turn missing after audio answer"
+        assert learner_turns[0]["text"] == "the fix was in the retry loop"
+
+    def test_empty_audio_is_422_not_500(self, client) -> None:
+        """Zero-byte upload must 422 before the provider call (a real
+        provider would raise the same way the mock does — validate first)."""
+        defense_id = _start(client)["defense_id"]
+        resp = client.post(
+            f"/v1/defense/{defense_id}/answer",
+            files={"audio": ("answer.wav", b"", "audio/wav")},
+        )
+        assert resp.status_code == 422, resp.text
+        stored = client.get(f"/v1/defense/{defense_id}").json()
+        assert len(stored["turns"]) == 1  # nothing appended
+
+    def test_answer_after_finish_is_409(self, client) -> None:
+        """A sealed transcript is append-only-no-more: the endpoints own
+        turn-vs-finalize sequencing (defense_store contract)."""
+        defense_id = _start(client)["defense_id"]
+        client.post(f"/v1/defense/{defense_id}/answer", data={"text": "a"})
+        assert client.post(f"/v1/defense/{defense_id}/finish").status_code == 200
+        resp = client.post(f"/v1/defense/{defense_id}/answer", data={"text": "late"})
+        assert resp.status_code == 409, resp.text
+        stored = client.get(f"/v1/defense/{defense_id}").json()
+        assert len(stored["turns"]) == 3  # ex, lrn, ex — no post-finish turns
+
+    def test_neither_text_nor_audio_422(self, client) -> None:
+        defense_id = _start(client)["defense_id"]
+        resp = client.post(f"/v1/defense/{defense_id}/answer")
+        assert resp.status_code == 422
+
+    def test_unknown_defense_404(self, client) -> None:
+        resp = client.post("/v1/defense/dfn-nope/answer", data={"text": "hi"})
+        assert resp.status_code == 404
+
+
+class TestAudioEndpoint:
+    def test_examiner_turn_streams_wav(self, client) -> None:
+        defense_id = _start(client)["defense_id"]
+        resp = client.get(f"/v1/defense/{defense_id}/audio/0")
+        assert resp.status_code == 200
+        assert resp.content
+        assert resp.headers["content-type"].startswith("audio/")
+
+    def test_unknown_turn_404(self, client) -> None:
+        defense_id = _start(client)["defense_id"]
+        assert client.get(f"/v1/defense/{defense_id}/audio/42").status_code == 404
+
+
+class TestFinishAndGet:
+    def test_full_loop_verdict_and_signals(self, client) -> None:
+        defense_id = _start(client)["defense_id"]
+        client.post(f"/v1/defense/{defense_id}/answer", data={"text": "answer one"})
+        finish = client.post(f"/v1/defense/{defense_id}/finish")
+        assert finish.status_code == 200, finish.text
+        body = finish.json()
+        assert body["verdict"]["verdict"] == "developing"
+        assert body["integrity_signals"]["pause_threshold_ms"]
+        stored = client.get(f"/v1/defense/{defense_id}").json()
+        assert stored["status"] == "finished"
+        assert stored["integrity_signals"]
+        # Must-Have #1: "verdict + transcript persisted" — the verdict must
+        # be retrievable from GET after finish, not only in the finish body.
+        assert stored["integrity_signals"]["verdict"]["verdict"] == "developing"
+
+    def test_long_pause_flagged(self, client, app) -> None:
+        from datetime import UTC, datetime
+
+        from ai_service.voice.defense_store import DefenseTurn
+
+        defense_id = _start(client)["defense_id"]
+        # inject a slow learner turn directly (simulated latency)
+        store = app.state.defense_store
+        store.append_turn(
+            defense_id,
+            DefenseTurn(
+                defense_id=defense_id,
+                seq=99,
+                role="learner",
+                text="slow reply",
+                ts=datetime.now(UTC),
+                latency_ms=30_000,
+                created_at=datetime.now(UTC),
+            ),
+        )
+        finish = client.post(f"/v1/defense/{defense_id}/finish")
+        assert finish.status_code == 200
+        signals = finish.json()["integrity_signals"]
+        assert any(p["turn"] == 99 for p in signals["long_pauses"])
+
+    def test_unknown_defense_404_on_all(self, client) -> None:
+        assert client.post("/v1/defense/dfn-nope/finish").status_code == 404
+        assert client.get("/v1/defense/dfn-nope").status_code == 404

apps/ai-service/tests/api/test_e2e_credential_flow.py

@@ -0,0 +1,216 @@
+"""Full credential-flow E2E over real engines (Task 6-5-01, REQ-3-007/008).
+
+Endpoint-level end-to-end with mock LLM/voice providers (G-2 precedent:
+real engine plumbing over real endpoints; provider choice is
+service-internal): variant -> telemetry-wired sandbox -> real in-sandbox
+exec -> trace -> grade -> oral defense -> verdict/signals -> proctor.
+No corpus fixture anywhere in the flow.
+
+Probe-guarded for user namespaces (the in-sandbox exec needs them).
+"""
+
+from __future__ import annotations
+
+import asyncio
+import contextlib
+import json
+import socket
+import time
+from pathlib import Path
+
+import pytest
+import uvicorn
+
+from ai_service.config import Settings
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+from ai_service.variants.store import SQLiteVariantStore
+from ai_service.voice.defense_store import SQLiteDefenseStore
+from tests.sandbox.test_isolation import USERSNS_AVAILABLE
+
+COACHING_JSON = json.dumps(
+    {
+        "summary": "Strong iteration.",
+        "strengths": ["Tests early."],
+        "gaps": ["One edge case missing."],
+        "next_steps": ["Add it."],
+    }
+)
+VERDICT_JSON = json.dumps(
+    {
+        "verdict": "developing",
+        "understanding": "Explains the build clearly.",
+        "process_justification": "Choices defended.",
+        "communication": "Clear.",
+        "strengths": ["Grounded in the digest."],
+        "gaps": ["Missed one edge case."],
+    }
+)
+PROCTOR_JSON = json.dumps(
+    {
+        "signals": [
+            {"signal_type": "idle_gap", "severity": "low", "note": "A short pause."}
+        ],
+        "intervention": "Keep momentum.",
+        "summary": "Healthy session.",
+    }
+)
+
+
+class FlowLLM(MockProvider):
+    """Prompt-discriminated: verdict vs coaching vs question vs proctor JSON."""
+
+    def _reply_for(self, messages, response_format):
+        all_text = "\n".join(m.content for m in messages)
+        if response_format is not None and response_format.get("type") == "json_object":
+            if "final verdict JSON" in all_text:
+                return VERDICT_JSON
+            if "rubric" in all_text.lower() and "Score this build session" in all_text:
+                return json.dumps(
+                    {
+                        "criteria": {
+                            "process_quality": 4,
+                            "correctness": 3,
+                            "debugging_discipline": 3,
+                            "test_usage": 4,
+                        },
+                        "strengths": ["Iterated with tests."],
+                        "gaps": ["One edge case missing."],
+                        "verdict": "developing",
+                    }
+                )
+            if "Explain it as coaching" in all_text:
+                return COACHING_JSON
+            if "integrity signals supportively" in all_text:
+                return PROCTOR_JSON
+            return json.dumps(
+                {
+                    "statement": (
+                        "Build a judge for code-review answers scoring factual "
+                        "accuracy with 3 edge cases and 5 test examples."
+                    )
+                }
+            )
+        return "Walk me through your last fix — what changed and why?"
+
+
+def _free_port() -> int:
+    with socket.socket() as s:
+        s.bind(("127.0.0.1", 0))
+        return s.getsockname()[1]
+
+
+@pytest.mark.asyncio
+async def test_full_credential_flow(tmp_path: Path) -> None:
+    if not USERSNS_AVAILABLE:
+        pytest.skip("user namespaces unavailable on this host (probe)")
+
+    import httpx
+
+    port = _free_port()
+    settings = Settings(provider="mock", voice_provider="mock", port=port)
+    app = create_app(settings)
+    app.state.provider = FlowLLM()
+    app.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+    app.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+    app.state.variant_store = SQLiteVariantStore(db_path=tmp_path / "v.db")
+    app.state.defense_store = SQLiteDefenseStore(db_path=tmp_path / "d.db")
+    app.state.trace_integrity = TraceIntegrityMap()
+
+    server = uvicorn.Server(uvicorn.Config(app, host="127.0.0.1", port=port, log_level="warning"))
+    serve_task = asyncio.get_running_loop().create_task(server.serve())
+    try:
+        for _ in range(100):
+            if server.started:
+                break
+            await asyncio.sleep(0.1)
+        assert server.started
+
+        base = f"http://127.0.0.1:{port}"
+        async with httpx.AsyncClient(base_url=base, timeout=30.0) as client:
+            # 1. Variant (real seeded generation, mock-rendered).
+            var = (await client.post("/v1/variants", json={
+                "learner_id": "pilot-learner", "competency_id": "stack-orchestration-c007",
+            })).json()
+            assert var["task_id"] and var["statement"] and var["starter_files"]
+            task_id = var["task_id"]
+
+            # 2. Telemetry-wired sandbox (real namespaces; agent joins via loopback).
+            sbx = (await client.post("/v1/sandboxes", json={
+                "learner_id": "pilot-learner", "task_id": task_id,
+            }))
+            assert sbx.status_code == 201, sbx.text
+            sandbox_id = sbx.json()["id"]
+
+            # 3. Real in-sandbox exec: run the starter test (capture agent streams
+            #    the workspace effects into the trace).
+            exec_resp = await client.post(f"/v1/sandboxes/{sandbox_id}/exec", json={
+                "cmd": ["pytest", "-q"],
+            })
+            assert exec_resp.status_code == 200, exec_resp.text
+
+            # 4. Trace: events landed in order (the capture agent runs async).
+            trace: list = []
+            deadline = time.monotonic() + 20.0
+            while time.monotonic() < deadline:
+                tr = await client.get(f"/v1/telemetry/traces/pilot-learner/{task_id}")
+                if tr.status_code == 200:
+                    trace = tr.json().get("events", [])
+                    if trace:
+                        break
+                await asyncio.sleep(0.25)
+            assert trace, "no telemetry events arrived from the real sandbox"
+            seqs = [e["seq"] for e in trace]
+            assert seqs == sorted(seqs)
+
+            # 5. Grade: rubric from the real digest (G-4 gate passed: no gaps).
+            grade = (await client.post("/v1/assessment/grade", json={
+                "learner_id": "pilot-learner", "task_id": task_id,
+            })).json()
+            assert grade["verdict"] == "GRADED", grade
+            assert grade["scores"]["criteria"]["process_quality"] == 4
+            assert grade["variant_seed"] == var["seed"]  # D-029 stamped
+
+            # 6. Assessor coaching FROM the stored grade.
+            coaching = (await client.post("/v1/assessment/evaluate", json={
+                "learner_id": "pilot-learner", "task_id": task_id,
+            })).json()
+            assert coaching["coaching"]["summary"]
+
+            # 7. Oral defense: start -> typed answers -> finish (mock voice).
+            defense = (await client.post("/v1/defense/start", json={
+                "learner_id": "pilot-learner", "task_id": task_id,
+            })).json()
+            assert defense["first_question"]
+            did = defense["defense_id"]
+            ans = await client.post(f"/v1/defense/{did}/answer", data={"text": "I fixed the loop."})
+            assert ans.status_code == 200, ans.text
+            finish = (await client.post(f"/v1/defense/{did}/finish")).json()
+            assert finish["verdict"]["verdict"] == "developing"
+            assert finish["integrity_signals"]
+
+            # 8. Proctor over the real digest + defense signals.
+            proctor = (await client.post("/v1/proctor/signals", json={
+                "learner_id": "pilot-learner", "task_id": task_id,
+            })).json()
+            assert proctor["intervention"]
+
+            # 9. Sandbox destroyed; no leaks.
+            destroy = await client.delete(f"/v1/sandboxes/{sandbox_id}")
+            assert destroy.status_code == 204
+            listed = (await client.get("/v1/sandboxes")).json()
+            assert all(s["id"] != sandbox_id for s in (listed.get("sandboxes") or []))
+
+            # 10. No corpus fixtures anywhere in this flow's payloads.
+            corpus_markers = ("lab-scenario", "proctor-scenario", "artifact-")
+            for payload in (var, grade, defense, finish, proctor):
+                assert not any(
+                    m in json.dumps(payload) for m in corpus_markers
+                ), "corpus fixture leaked into the learner path"
+    finally:
+        server.should_exit = True
+        with contextlib.suppress(Exception):
+            await asyncio.wait_for(serve_task, timeout=10.0)

apps/ai-service/tests/api/test_grading.py

@@ -0,0 +1,397 @@
+"""Assessment grade endpoint tests — grading engine over HTTP (Task 3-3-01).
+
+Contract under test (api/assessment.py, REQ-3-004):
+
+    POST /v1/assessment/grade {learner_id, task_id}
+        GRADED                    → 200, rubric scores + digest summary
+        UNGRADABLE_TRACE_INCOMPLETE → 200, gate record (missing_seqs /
+                                    integrity_flag surfaced in scores)
+        UNGRADABLE_EMPTY_TRACE    → 200, gate record (documented choice: an
+                                    unknown task is ALSO an empty pair; the
+                                    engine cannot distinguish, and the gate
+                                    outcome is a durable first-class result)
+        StructuredOutputError     → 502 (provider exhausted the D-020 budget)
+    GET  /v1/assessment/grade/{learner_id}/{task_id}
+        stored latest grade       → 200 (same scores as the POST)
+        never-graded pair         → 404
+
+Wiring: per-test tmp-path SQLite stores + a pre-set GradingEngine
+(state-injection override — the lifespan adopts trace_store/grade_store/
+trace_integrity/grading_engine from app.state instead of constructing
+them; same pattern as test_telemetry_ingest.py / test_sandboxes.py). The
+engine binds a scripted provider so each test controls the LLM exactly,
+and counts calls so gate tests can assert the LLM was never reached
+(G-4 holds through the whole HTTP stack).
+
+Zero network: providers are MockProvider subclasses only (conftest rule).
+"""
+
+from __future__ import annotations
+
+from collections.abc import Iterator
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.config import Settings
+from ai_service.corpus.trace_fixtures import STRONG_BUILDER
+from ai_service.grading.engine import GradingEngine
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.mock import MockProvider, ScriptedJSONProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.models import TelemetryEvent
+from ai_service.telemetry.store import SQLiteTraceStore
+
+LEARNER = "grade-learner"
+TASK = "task-grade-1"
+T0 = datetime(2026, 9, 12, 3, 0, 0, tzinfo=UTC)
+
+#: Canonical well-formed rubric payload (matches grading.engine.RubricScore).
+RUBRIC_PAYLOAD: dict = {
+    "criteria": {
+        "process_quality": 4,
+        "correctness": 4,
+        "debugging_discipline": 3,
+        "test_usage": 4,
+    },
+    "strengths": ["tight edit-test loops throughout"],
+    "gaps": ["final commit discipline loose"],
+    "verdict": "mastered",
+}
+
+
+class CountingScriptedProvider(ScriptedJSONProvider):
+    """ScriptedJSONProvider that counts chat() calls.
+
+    Gate tests assert calls == 0 (the LLM is never reached through the
+    whole HTTP stack — G-4); happy-path tests sanity-check calls >= 1.
+    """
+
+    def __init__(self, payload: dict) -> None:
+        super().__init__(payload)
+        self.calls = 0
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):
+        self.calls += 1
+        return await super().chat(
+            messages, model=model, temperature=temperature, response_format=response_format
+        )
+
+
+def _event(seq: int, kind: str, payload: dict | None = None, offset_s: float = 0.0):
+    return TelemetryEvent(
+        learner_id=LEARNER,
+        task_id=TASK,
+        seq=seq,
+        kind=kind,
+        payload=payload or {},
+        ts=T0 + timedelta(seconds=offset_s),
+        sandbox_id="sbx-grade",
+    )
+
+
+def _complete_trace() -> list[TelemetryEvent]:
+    """Contiguous seq 0..6 — a gradeable trace (gap-free, unflagged)."""
+    return [
+        _event(0, "activity", {"state": "starting"}, 0.0),
+        _event(1, "file_diff", {"path": "a.py", "added": 12}, 10.0),
+        _event(2, "command", {"cmd": "pytest -q"}, 20.0),
+        _event(3, "test_result", {"passed": False, "exit_code": 1}, 25.0),
+        _event(4, "file_diff", {"path": "a.py", "added": 4, "removed": 2}, 40.0),
+        _event(5, "command", {"cmd": "pytest -q"}, 60.0),
+        _event(6, "test_result", {"passed": True, "exit_code": 0}, 65.0),
+    ]
+
+
+@pytest.fixture()
+def trace_store(tmp_path: Path) -> Iterator[SQLiteTraceStore]:
+    store = SQLiteTraceStore(db_path=tmp_path / "traces.db")
+    yield store
+    store.close()
+
+
+@pytest.fixture()
+def grade_store(tmp_path: Path) -> Iterator[SQLiteGradeStore]:
+    store = SQLiteGradeStore(db_path=tmp_path / "grades.db")
+    yield store
+    store.close()
+
+
+@pytest.fixture()
+def integrity() -> TraceIntegrityMap:
+    return TraceIntegrityMap()
+
+
+def _make_client(
+    tmp_path: Path,
+    trace_store: SQLiteTraceStore,
+    grade_store: SQLiteGradeStore,
+    integrity: TraceIntegrityMap,
+    provider: MockProvider,
+) -> TestClient:
+    """App + TestClient with stores, integrity map and a pre-set engine.
+
+    The lifespan adopts every pre-set service (state-injection override);
+    the engine binds OUR provider, so the fixture — not Settings — scripts
+    the LLM. Cloud-free guard: the provider must be a MockProvider family
+    member (conftest rule, enforced here because this module builds its
+    own client rather than consuming the conftest one).
+    """
+    assert isinstance(provider, MockProvider)
+    settings = Settings(
+        provider="mock",
+        db_path=tmp_path / "grading-test.db",
+        sandbox_dir=tmp_path / "sandboxes",
+    )
+    app = create_app(settings)
+    app.state.trace_store = trace_store
+    app.state.grade_store = grade_store
+    app.state.trace_integrity = integrity
+    app.state.grading_engine = GradingEngine(
+        trace_store, grade_store, integrity, provider, model="gemma4:31b"
+    )
+    return TestClient(app)
+
+
+def _seed(store: SQLiteTraceStore, events: list[TelemetryEvent]) -> None:
+    for e in events:
+        store.append(e)
+
+
+def _post(client: TestClient, learner: str = LEARNER, task: str = TASK):
+    return client.post("/v1/assessment/grade", json={"learner_id": learner, "task_id": task})
+
+
+def _get(client: TestClient, learner: str = LEARNER, task: str = TASK):
+    return client.get(f"/v1/assessment/grade/{learner}/{task}")
+
+
+# -- happy path: complete trace → GRADED -----------------------------------------
+
+
+class TestGraded:
+    def test_post_complete_trace_returns_rubric_scores_and_digest(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        _seed(trace_store, _complete_trace())
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _post(client)
+
+        assert response.status_code == 200
+        body = response.json()
+        assert body["verdict"] == "GRADED"
+        assert body["learner_id"] == LEARNER
+        assert body["task_id"] == TASK
+        assert body["scores"] == RUBRIC_PAYLOAD  # rubric verdict rides in scores
+        assert body["scores"]["verdict"] == "mastered"
+        assert body["variant_seed"] is None  # null until P4 (D-029)
+        assert body["model"] == "gemma4:31b"  # provenance travels
+        assert provider.calls == 1  # happy path: exactly one LLM call
+        # digest summary rides along (D-028 reproducible input)
+        assert body["digest"]["event_count"] == 7
+        assert body["digest"]["final_test_status"] == "pass"
+
+    def test_corpus_fixture_trace_grades(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        """The strong-builder corpus fixture (real calibration trace) is
+        gradeable over HTTP end to end."""
+        _seed(trace_store, STRONG_BUILDER.events)
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _post(
+                client, learner=STRONG_BUILDER.learner_id, task=STRONG_BUILDER.task_id
+            )
+
+        assert response.status_code == 200
+        body = response.json()
+        assert body["verdict"] == "GRADED"
+        assert body["digest"]["event_count"] == len(STRONG_BUILDER.event_specs)
+
+    def test_get_after_post_returns_same_stored_scores(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        _seed(trace_store, _complete_trace())
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            posted = _post(client)
+            assert posted.status_code == 200
+
+            fetched = _get(client)
+            assert fetched.status_code == 200
+            stored = fetched.json()
+
+        assert stored["scores"] == posted.json()["scores"]
+        assert stored["verdict"] == "GRADED"
+        assert stored["digest"] == posted.json()["digest"]
+        assert stored["created_at"] == posted.json()["created_at"]
+
+    def test_post_regrade_upserts_get_returns_latest(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        """POST twice → one row holding the LATEST grade (GradeStore upsert)."""
+        _seed(trace_store, _complete_trace())
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            first = _post(client)
+            assert first.status_code == 200
+            assert first.json()["scores"]["verdict"] == "mastered"
+
+            # the same provider now scripts a different (weaker) rubric
+            updated = {
+                **RUBRIC_PAYLOAD,
+                "criteria": {**RUBRIC_PAYLOAD["criteria"], "process_quality": 1},
+                "verdict": "not_yet",
+            }
+            provider.payload = updated
+            second = _post(client)
+            assert second.status_code == 200
+            assert second.json()["scores"] == updated
+
+            # GET returns the LATEST grade, not the first
+            fetched = _get(client)
+            assert fetched.json()["scores"] == updated
+            assert fetched.json()["scores"]["verdict"] == "not_yet"
+
+        # one row, latest wins (upsert, not append)
+        grades = grade_store.list_for_learner(LEARNER)
+        assert len(grades) == 1
+        assert grades[0].scores == updated
+
+
+# -- G-4 gate outcomes over HTTP: 200 with the ungradable record -------------------
+
+
+class TestGateOutcomes:
+    def test_gapped_trace_200_ungradable_incomplete_with_missing_seqs(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        _seed(trace_store, [e for e in _complete_trace() if e.seq != 2])  # seq 2 missing
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _post(client)
+
+        assert response.status_code == 200  # ungradable IS a valid result — not 5xx
+        body = response.json()
+        assert body["verdict"] == "UNGRADABLE_TRACE_INCOMPLETE"
+        assert body["scores"]["missing_seqs"] == [2]
+        assert body["scores"]["integrity_flag"] is None
+        assert body["digest"] == {}  # nothing was graded
+        assert body["model"] == "none"  # no LLM involved (honest provenance)
+        assert provider.calls == 0, "LLM was called despite a gapped trace (G-4)"
+
+    def test_flooded_trace_200_ungradable_incomplete_with_integrity_reason(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        """Integrity-flagged trace (G-3 INCOMPLETE_FLOODED) surfaces the flag
+        reason in scores — the same verdict as gaps, different detail."""
+        _seed(trace_store, _complete_trace())
+        integrity.mark(LEARNER, TASK, "INCOMPLETE_FLOODED")
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _post(client)
+
+        assert response.status_code == 200
+        body = response.json()
+        assert body["verdict"] == "UNGRADABLE_TRACE_INCOMPLETE"
+        assert body["scores"]["integrity_flag"] == "INCOMPLETE_FLOODED"
+        assert body["scores"]["missing_seqs"] == []  # rows complete but untrusted
+        assert provider.calls == 0, "LLM was called despite an integrity flag (G-4)"
+
+    def test_empty_trace_200_ungradable_empty(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        """Zero events for the pair (including a task that never had a
+        trace) → UNGRADABLE_EMPTY_TRACE, persisted like any grade."""
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _post(client, learner="ghost-learner", task="never-started")
+
+        assert response.status_code == 200
+        body = response.json()
+        assert body["verdict"] == "UNGRADABLE_EMPTY_TRACE"
+        assert body["scores"] == {"integrity_flag": None, "missing_seqs": []}
+        assert body["digest"] == {}
+        assert body["model"] == "none"
+        assert provider.calls == 0
+
+        # the gate record is durable: GET now returns it (no longer 404)
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            fetched = _get(client, learner="ghost-learner", task="never-started")
+        assert fetched.status_code == 200
+        assert fetched.json()["verdict"] == "UNGRADABLE_EMPTY_TRACE"
+
+
+# -- provider failure → 502 --------------------------------------------------------
+
+
+class TestProviderFailure:
+    def test_persistent_malformed_llm_output_502(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        """Stock MockProvider: its json_object reply is wrong-shaped, so the
+        D-020 defense exhausts both attempts → endpoint maps to 502 and
+        nothing is persisted."""
+        _seed(trace_store, _complete_trace())
+        provider = MockProvider()  # always malformed for the rubric schema
+
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _post(client)
+
+        assert response.status_code == 502
+        assert "grading failed" in response.json()["detail"]
+        # no fabricated/partial record was persisted
+        assert grade_store.get(LEARNER, TASK) is None
+
+    def test_502_leaves_earlier_grade_intact(
+        self, tmp_path, trace_store, grade_store, integrity
+    ):
+        """A failed regrade must not clobber the previously stored grade."""
+        _seed(trace_store, _complete_trace())
+        scripted = CountingScriptedProvider(RUBRIC_PAYLOAD)
+        with _make_client(tmp_path, trace_store, grade_store, integrity, scripted) as client:
+            assert _post(client).status_code == 200
+
+        # regrade attempt hits a provider that now always fails
+        broken = MockProvider()
+        with _make_client(tmp_path, trace_store, grade_store, integrity, broken) as client:
+            assert _post(client).status_code == 502
+            fetched = _get(client)
+
+        assert fetched.status_code == 200  # earlier grade still readable
+        assert fetched.json()["scores"] == RUBRIC_PAYLOAD
+
+
+# -- stored-grade reads ------------------------------------------------------------
+
+
+class TestStoredGradeReads:
+    def test_get_unknown_pair_404(self, tmp_path, trace_store, grade_store, integrity):
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            response = _get(client, learner="nobody", task="never-graded")
+
+        assert response.status_code == 404
+        assert "no stored grade" in response.json()["detail"]
+
+    def test_missing_body_fields_422(self, tmp_path, trace_store, grade_store, integrity):
+        provider = CountingScriptedProvider(RUBRIC_PAYLOAD)
+        with _make_client(tmp_path, trace_store, grade_store, integrity, provider) as client:
+            assert client.post("/v1/assessment/grade", json={}).status_code == 422
+            assert (
+                client.post(
+                    "/v1/assessment/grade", json={"learner_id": LEARNER}
+                ).status_code
+                == 422
+            )

apps/ai-service/tests/api/test_lab.py

@@ -1,47 +1,86 @@
-"""Lab feedback endpoint tests — SSE envelope with agent=lab (REQ-2-007)."""
+"""Lab endpoint tests — LIVE trace contract (REQ-3-007).
 
-import json
+v0.3 re-grounding: POST /v1/lab/feedback takes {learner_id, task_id}; the
+digest is computed from the learner's real TraceStore events. No corpus
+scenarios; empty trace is a valid "no telemetry yet" coaching path.
+"""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.config import Settings
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.models import TelemetryEvent
+from ai_service.telemetry.store import SQLiteTraceStore
+
+T0 = datetime(2026, 9, 12, tzinfo=UTC)
 
 
-def stream_events(client, payload) -> list[dict]:
-    with client.stream("POST", "/v1/lab/feedback", json=payload) as response:
-        assert response.status_code == 200
-        events = []
-        for line in response.iter_lines():
-            if line.startswith("data:"):
-                d = line.removeprefix("data:").strip()
-                if d == "[DONE]":
-                    events.append({"type": "[DONE]"})
-                else:
-                    events.append(json.loads(d))
-        return events
+def _event(seq: int, kind: str, payload: dict, offset_s: float):
+    return TelemetryEvent(
+        learner_id="lab-learner",
+        task_id="lab-task",
+        seq=seq,
+        kind=kind,
+        payload=payload,
+        ts=T0 + timedelta(seconds=offset_s),
+        sandbox_id="sbx-lab",
+    )
 
 
-def test_lab_feedback_streams_full_envelope(client):
-    events = stream_events(client, {"scenario_id": "lab-scenario-strong"})
-    assert events[0]["type"] == "meta"
-    assert events[0]["agent"] == "lab"
-    assert events[0]["scenario_id"] == "lab-scenario-strong"
-    deltas = [e for e in events if e["type"] == "delta"]
-    assert len(deltas) >= 1
-    assert any(e["type"] == "done" for e in events)
-    assert events[-1]["type"] == "[DONE]"
+class StreamingMock(MockProvider):
+    """Deterministic token stream for the SSE path."""
+
+    def _reply_for(self, messages, response_format):
+        return "Feedback grounded in your live session digest."
 
 
-def test_unknown_scenario_404(client):
-    response = client.post("/v1/lab/feedback", json={"scenario_id": "nope"})
-    assert response.status_code == 404
-    assert "nope" in response.json()["detail"]
+@pytest.fixture()
+def client(tmp_path: Path) -> TestClient:
+    app = create_app(Settings(provider="mock"))
+    app.state.provider = StreamingMock()
+    app.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+    app.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+    app.state.trace_integrity = TraceIntegrityMap()
+    with TestClient(app) as c:
+        yield c
 
 
-def test_distinct_scenarios_distinct_replies(client):
-    strong = stream_events(client, {"scenario_id": "lab-scenario-strong"})
-    struggling = stream_events(client, {"scenario_id": "lab-scenario-struggling"})
-    strong_text = "".join(e["content"] for e in strong if e["type"] == "delta")
-    struggling_text = "".join(e["content"] for e in struggling if e["type"] == "delta")
-    assert strong_text != struggling_text
+def test_live_trace_streams_full_envelope(client: TestClient) -> None:
+    store: SQLiteTraceStore = client.app.state.trace_store
+    for e in [
+        _event(0, "file_diff", {"path": "x.py"}, 0),
+        _event(1, "command", {"cmd": "pytest -q"}, 10),
+        _event(2, "test_result", {"passed": False, "exit_code": 1}, 20),
+        _event(3, "test_result", {"passed": True, "exit_code": 0}, 40),
+    ]:
+        store.append(e)
+    with client.stream(
+        "POST", "/v1/lab/feedback", json={"learner_id": "lab-learner", "task_id": "lab-task"}
+    ) as resp:
+        assert resp.status_code == 200
+        body = "".join(chunk.decode() for chunk in resp.iter_raw())
+    assert '"agent": "lab"' in body or '"agent":"lab"' in body
+    assert '"task_id": "lab-task"' in body
+    assert '"type": "delta"' in body or '"type":"delta"' in body
 
 
-def test_missing_scenario_id_422(client):
-    response = client.post("/v1/lab/feedback", json={})
-    assert response.status_code == 422
+def test_empty_trace_coaches_the_baseline(client: TestClient) -> None:
+    """No telemetry is NOT an error — Lab coaches 'run the starter test'."""
+    with client.stream(
+        "POST", "/v1/lab/feedback", json={"learner_id": "lab-learner", "task_id": "no-events"}
+    ) as resp:
+        assert resp.status_code == 200
+
+
+def test_missing_fields_422(client: TestClient) -> None:
+    resp = client.post("/v1/lab/feedback", json={"learner_id": "x"})
+    assert resp.status_code == 422

apps/ai-service/tests/api/test_proctor.py

@@ -1,49 +1,121 @@
-"""Proctor signals endpoint tests — validated JSON, 404s (REQ-2-009)."""
+"""Proctor signals endpoint tests — REAL inputs contract (REQ-3-007).
 
-from ai_service.agents.proctor import ProctorAssessment
-from ai_service.llm.mock import ScriptedJSONProvider
+v0.3 re-grounding: POST /v1/proctor/signals takes {learner_id, task_id}
+and gathers digest + defense signals + variant context server-side.
+"""
+
+from __future__ import annotations
+
+import json
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.config import Settings
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.models import TelemetryEvent
+from ai_service.telemetry.store import SQLiteTraceStore
+from ai_service.variants.store import SQLiteVariantStore
+from ai_service.voice.defense_store import SQLiteDefenseStore
 
 VALID = {
-    "scenario_id": "proctor-scenario-distracted",
     "signals": [
-        {"signal_type": "context_switch", "severity": "low",
-         "note": "Docs tab at t+120s is normal"},
-        {"signal_type": "idle_gap", "severity": "medium",
-         "note": "5-minute idle at t+300s"},
+        {"signal_type": "idle_gap", "severity": "low", "note": "One long pause."},
     ],
     "intervention": "Offer a short break, then restate the plan",
     "summary": "Coaching-shaped session note",
 }
 
-
-def test_signals_returns_validated_json(client):
-    original = client.app.state.provider
-    client.app.state.provider = ScriptedJSONProvider(VALID)
-    try:
-        response = client.post(
-            "/v1/proctor/signals", json={"scenario_id": "proctor-scenario-distracted"}
-        )
-    finally:
-        client.app.state.provider = original
-    assert response.status_code == 200
-    validated = ProctorAssessment.model_validate(response.json())
-    assert validated.scenario_id == "proctor-scenario-distracted"
-    assert validated.intervention
+T0 = datetime(2026, 9, 12, tzinfo=UTC)
 
 
-def test_unknown_scenario_404(client):
-    response = client.post("/v1/proctor/signals", json={"scenario_id": "ghost"})
-    assert response.status_code == 404
+class ProctorJSON(MockProvider):
+    def _reply_for(self, messages, response_format):
+        if response_format is not None and response_format.get("type") == "json_object":
+            return json.dumps(VALID)
+        return super()._reply_for(messages, response_format)
 
 
-def test_unparseable_provider_502(client):
-    response = client.post(
-        "/v1/proctor/signals", json={"scenario_id": "proctor-scenario-healthy"}
+class BrokenJSON(MockProvider):
+    def _reply_for(self, messages, response_format):
+        if response_format is not None and response_format.get("type") == "json_object":
+            return "not json ever"
+        return super()._reply_for(messages, response_format)
+
+
+@pytest.fixture()
+def client(tmp_path: Path) -> TestClient:
+    app = create_app(Settings(provider="mock"))
+    app.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+    app.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+    app.state.variant_store = SQLiteVariantStore(db_path=tmp_path / "v.db")
+    app.state.defense_store = SQLiteDefenseStore(db_path=tmp_path / "d.db")
+    app.state.trace_integrity = TraceIntegrityMap()
+    with TestClient(app) as c:
+        yield c
+
+
+def _seed_trace(client: TestClient) -> None:
+    store: SQLiteTraceStore = client.app.state.trace_store
+    for e in [
+        TelemetryEvent(
+            learner_id="p-learner",
+            task_id="p-task",
+            seq=0,
+            kind="activity",
+            payload={"state": "idle"},
+            ts=T0,
+            sandbox_id="sbx-p",
+        ),
+        TelemetryEvent(
+            learner_id="p-learner",
+            task_id="p-task",
+            seq=1,
+            kind="activity",
+            payload={"state": "idle"},
+            ts=T0 + timedelta(seconds=400),
+            sandbox_id="sbx-p",
+        ),
+    ]:
+        store.append(e)
+
+
+def test_signals_returns_validated_json(client: TestClient) -> None:
+    client.app.state.provider = ProctorJSON()
+    _seed_trace(client)
+    resp = client.post(
+        "/v1/proctor/signals", json={"learner_id": "p-learner", "task_id": "p-task"}
     )
-    assert response.status_code == 502
-    assert "failed" in response.json()["detail"].lower()
+    assert resp.status_code == 200, resp.text
+    body = resp.json()
+    assert body["intervention"]
+    assert body["signals"][0]["signal_type"] == "idle_gap"
 
 
-def test_missing_scenario_id_422(client):
-    response = client.post("/v1/proctor/signals", json={})
-    assert response.status_code == 422
+def test_empty_trace_is_valid_not_404(client: TestClient) -> None:
+    """No telemetry → the proctor still assesses (nothing to flag)."""
+    client.app.state.provider = ProctorJSON()
+    resp = client.post(
+        "/v1/proctor/signals", json={"learner_id": "nobody", "task_id": "nothing"}
+    )
+    assert resp.status_code == 200
+
+
+def test_unparseable_provider_502(client: TestClient) -> None:
+    client.app.state.provider = BrokenJSON()
+    _seed_trace(client)
+    resp = client.post(
+        "/v1/proctor/signals", json={"learner_id": "p-learner", "task_id": "p-task"}
+    )
+    assert resp.status_code == 502
+
+
+def test_missing_fields_422(client: TestClient) -> None:
+    client.app.state.provider = ProctorJSON()
+    resp = client.post("/v1/proctor/signals", json={"learner_id": "x"})
+    assert resp.status_code == 422

apps/ai-service/tests/api/test_sandboxes.py

@@ -359,3 +359,91 @@ def test_real_backend_create_path_runs(tmp_path: Path) -> None:
         assert (sandbox_root / sandbox_id / "workspace").is_dir()
     finally:
         shutil.rmtree(sandbox_root, ignore_errors=True)
+
+
+class TestFilesAndExecRoutes:
+    """Workspace CRUD + Run/Test exec (Phase 6, REQ-3-008, CUT-2)."""
+
+    def test_file_write_read_list_roundtrip(self, client):
+        handle = client.post(
+            "/v1/sandboxes", json={"learner_id": "pilot-learner"}
+        ).json()
+        sbx = handle["id"]
+        put = client.put(
+            f"/v1/sandboxes/{sbx}/files/main.py",
+            json={"path": "main.py", "content": "print('hi')"},
+        )
+        assert put.status_code == 200, put.text
+        got = client.get(f"/v1/sandboxes/{sbx}/files/main.py")
+        assert got.status_code == 200
+        assert "print('hi')" in got.json()["content"]
+        listed = client.get(f"/v1/sandboxes/{sbx}/files")
+        assert "main.py" in listed.json()["files"]
+        client.delete(f"/v1/sandboxes/{sbx}")
+
+    def test_traversal_rejected(self, client):
+        handle = client.post(
+            "/v1/sandboxes", json={"learner_id": "pilot-learner"}
+        ).json()
+        sbx = handle["id"]
+        bad = client.put(
+            f"/v1/sandboxes/{sbx}/files/..%2Fescape.txt",
+            json={"path": "../escape.txt", "content": "x"},
+        )
+        assert bad.status_code == 422
+        client.delete(f"/v1/sandboxes/{sbx}")
+
+    def test_unknown_file_404(self, client):
+        handle = client.post(
+            "/v1/sandboxes", json={"learner_id": "pilot-learner"}
+        ).json()
+        sbx = handle["id"]
+        assert client.get(f"/v1/sandboxes/{sbx}/files/ghost.py").status_code == 404
+        client.delete(f"/v1/sandboxes/{sbx}")
+
+    def test_exec_unknown_sandbox_404(self, client):
+        resp = client.post(
+            "/v1/sandboxes/sbx-nope/exec", json={"cmd": ["echo", "hi"]}
+        )
+        assert resp.status_code == 404
+
+    def test_unknown_sandbox_file_routes_404_not_500(self, client):
+        """P7: read/write on an unknown sandbox must 404 (SandboxNotFoundError
+        previously escaped _workspace_dir as an unhandled 500)."""
+        assert (
+            client.get("/v1/sandboxes/sbx-nope/files/whatever.py").status_code == 404
+        )
+        put = client.put(
+            "/v1/sandboxes/sbx-nope/files/whatever.py",
+            json={"path": "whatever.py", "content": "x"},
+        )
+        assert put.status_code == 404
+
+    def test_symlink_escape_rejected(self, client):
+        """P7: an exec-planted symlink in the workspace must not let the
+        file routes read/write OUTSIDE the bind (lexical traversal checks
+        cannot see symlinks — resolve + containment re-check is the gate)."""
+        handle = client.post(
+            "/v1/sandboxes", json={"learner_id": "pilot-learner"}
+        ).json()
+        sbx = handle["id"]
+        workspace = Path(handle["workdir"]) / "workspace"
+        outside = workspace.parent / "secret.txt"
+        outside.write_text("host secret")  # a host file OUTSIDE the bind
+        try:
+            (workspace / "leak.txt").symlink_to(outside)
+            read = client.get(f"/v1/sandboxes/{sbx}/files/leak.txt")
+            assert read.status_code == 422, (
+                f"symlink escape read must 422, got {read.status_code}: {read.text}"
+            )
+            write = client.put(
+                f"/v1/sandboxes/{sbx}/files/leak.txt",
+                json={"path": "leak.txt", "content": "pwned"},
+            )
+            assert write.status_code == 422, (
+                f"symlink escape write must 422, got {write.status_code}: {write.text}"
+            )
+            assert outside.read_text() == "host secret"  # untouched
+        finally:
+            client.delete(f"/v1/sandboxes/{sbx}")
+            outside.unlink(missing_ok=True)

apps/ai-service/tests/api/test_telemetry_ingest.py

@@ -0,0 +1,451 @@
+"""WS telemetry ingest API tests (REQ-3-003, D-026, G-3).
+
+Frame contract under test (telemetry/ingest.py):
+
+    WS /v1/telemetry/ingest?learner_id=L&task_id=T[&sandbox_id=S]
+    client → server: one event per JSON text frame
+                     {"seq", "kind", "payload", "ts", "sandbox_id"?}
+    server → client: {"type": "gap_warning"|"event_rejected"|"flooded"|"ack_total"}
+    close:  1000 clean flush · 1008 policy violation (G-3 flood)
+
+Each test runs against a per-test SQLite file in tmp_path via the same
+state-injection override the sandbox API tests use: build a real
+SQLiteTraceStore, pre-set `app.state.trace_store`, let the lifespan adopt it.
+
+Covered:
+    - connect → send 3 events → GET trace returns them ordered
+    - resend event 2 → deduped (no double-store; at-least-once contract)
+    - skip to seq 5 → gap_warning frame + GET /gaps reports missing seqs
+    - burst past a lowered cap → 1008 close + INCOMPLETE_FLOODED +
+      is_incomplete(...) true
+    - queue-overflow path → same 1008 + INCOMPLETE_FLOODED
+    - reconnect after flood still cannot append (flag is terminal)
+    - unknown trace GET → 404 (traces + gaps)
+    - missing identity query params → handshake rejected
+"""
+
+from __future__ import annotations
+
+import json
+import time
+from collections.abc import Iterator
+from datetime import UTC, datetime
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+from starlette.websockets import WebSocketDisconnect
+
+from ai_service.config import Settings
+from ai_service.main import create_app
+from ai_service.telemetry import ingest as ingest_mod
+from ai_service.telemetry.store import SQLiteTraceStore
+
+LEARNER = "learner-9"
+TASK = "task-9"
+_BASE_TS = datetime(2026, 9, 11, 12, 0, 0, tzinfo=UTC)
+
+KNOWN_KINDS = ("command", "file_diff", "run_result", "test_result", "activity")
+
+
+def _frame(seq: int, kind: str = "command", sandbox_id: str = "") -> str:
+    return json.dumps(
+        {
+            "seq": seq,
+            "kind": kind,
+            "payload": {"n": seq},
+            "ts": _BASE_TS.isoformat(),
+            **({"sandbox_id": sandbox_id} if sandbox_id else {}),
+        }
+    )
+
+
+def _drain(ws) -> None:
+    """Give the portal a beat to deliver queued frames.
+
+    Starlette's TestClient runs the app on a portal thread; frames produced in
+    the app loop still need a portal round-trip to reach the test thread. A
+    real (tiny) sleep flushes them deterministically without hardcoding
+    protocol timing.
+    """
+    time.sleep(0.05)
+
+
+def _recv_status(ws) -> dict:
+    """Next JSON status frame, skipping keepalive pings; re-raise close.
+
+    TestClient delivers close as a `websocket.close` message dict, not by
+    itself raising — `_recv_status` must surface it as WebSocketDisconnect so
+    `pytest.raises(WebSocketDisconnect)` around a read loop terminates.
+    """
+    while True:
+        msg = ws.receive()
+        if msg.get("bytes") is not None:  # keepalive ping — not a status frame
+            continue
+        if msg["type"] == "websocket.close":
+            raise WebSocketDisconnect(msg.get("code", 1000), msg.get("reason", ""))
+        return json.loads(msg["text"])
+
+
+def _ingest_url(learner_id: str = LEARNER, task_id: str = TASK, sandbox_id: str = "") -> str:
+    url = f"/v1/telemetry/ingest?learner_id={learner_id}&task_id={task_id}"
+    return f"{url}&sandbox_id={sandbox_id}" if sandbox_id else url
+
+
+@pytest.fixture()
+def store(tmp_path: Path) -> Iterator[SQLiteTraceStore]:
+    s = SQLiteTraceStore(db_path=tmp_path / "telemetry-test.db")
+    yield s
+    s.close()
+
+
+@pytest.fixture()
+def app(store: SQLiteTraceStore, tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> Iterator:
+    # Keep the keepalive interval comfortably above the per-test read window:
+    # a ping that fires mid-assertion interleaves with status frames and the
+    # reader must skip it; at 2s a single test's reads all land ping-free
+    # while a *blocked* read still fails fast on the next ping tick.
+    monkeypatch.setattr(ingest_mod, "PING_INTERVAL_S", 2.0)
+    settings = Settings(
+        provider="mock",
+        db_path=tmp_path / "telemetry-test.db",
+        sandbox_dir=tmp_path / "sandboxes",
+        telemetry_max_events_per_task=5,  # lowered G-3 cap for the flood test
+    )
+    application = create_app(settings)
+    application.state.trace_store = store  # state-injection override
+    yield application
+
+
+@pytest.fixture()
+def client(app) -> Iterator[TestClient]:
+    with TestClient(app) as c:
+        yield c
+
+
+# -- happy path: ordered trace retrieval --------------------------------------
+
+
+def test_connect_send_three_get_trace_ordered(client: TestClient) -> None:
+    with client.websocket_connect(_ingest_url()) as ws:
+        # Arrive out of arrival order? No — seq arrival order here; ordering is
+        # by seq at read. Send 0,1,2 then close cleanly.
+        for seq in (0, 1, 2):
+            ws.send_text(_frame(seq))
+
+    response = client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}")
+    assert response.status_code == 200
+    body = response.json()
+    assert [e["seq"] for e in body["events"]] == [0, 1, 2]
+    assert [e["kind"] for e in body["events"]] == ["command"] * 3
+    assert all(e["learner_id"] == LEARNER for e in body["events"])
+    assert body["incomplete"] is False
+    assert body["integrity_reason"] is None
+
+
+def test_different_kinds_and_sandbox_id_roundtrip(client: TestClient) -> None:
+    with client.websocket_connect(_ingest_url(sandbox_id="sbx-42")) as ws:
+        for seq, kind in enumerate(KNOWN_KINDS[:3]):
+            ws.send_text(_frame(seq, kind))
+
+    body = client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}").json()
+    assert [e["kind"] for e in body["events"]] == list(KNOWN_KINDS[:3])
+    # Frame had no sandbox_id → the connection's query-param sandbox_id applies.
+    assert all(e["sandbox_id"] == "sbx-42" for e in body["events"])
+
+
+# -- idempotent dedup (at-least-once, server-side) ------------------------------
+
+
+def test_resend_event_deduped_no_double_store(client: TestClient) -> None:
+    with client.websocket_connect(_ingest_url()) as ws:
+        for seq in (0, 1, 2):
+            ws.send_text(_frame(seq))
+        ws.send_text(_frame(1, kind="file_diff"))  # at-least-once retry of seq 1
+
+    body = client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}").json()
+    assert [e["seq"] for e in body["events"]] == [0, 1, 2]  # stored exactly once
+    # First write wins — the retry's different body must NOT overwrite.
+    assert body["events"][1]["kind"] == "command"
+
+
+# -- gap detection --------------------------------------------------------------
+
+
+def test_skip_to_seq_5_reports_gap_warning_and_gaps_endpoint(
+    client: TestClient,
+) -> None:
+    with client.websocket_connect(_ingest_url()) as ws:
+        for seq in (0, 1, 2):
+            ws.send_text(_frame(seq))
+        ws.send_text(_frame(5))  # skips 3,4 → gap_warning frame
+        _drain(ws)
+        frames = [_recv_status(ws)]
+    assert frames == [{"type": "gap_warning", "missing_seqs": [3, 4]}]
+
+    gaps = client.get(f"/v1/telemetry/gaps/{LEARNER}/{TASK}")
+    assert gaps.status_code == 200
+    assert gaps.json()["gaps"] == [3, 4]
+    assert gaps.json()["incomplete"] is False
+
+
+# -- G-3 flood control: cap breach → 1008 + INCOMPLETE_FLOODED --------------------
+
+
+def test_burst_past_cap_closes_1008_and_marks_incomplete_flooded(
+    client: TestClient, app, store: SQLiteTraceStore
+) -> None:
+    # Cap is 5 (fixture); the 6th event trips cap_exceeded.
+    with client.websocket_connect(_ingest_url()) as ws:
+        for seq in range(6):
+            ws.send_text(_frame(seq))
+        _drain(ws)  # let the app flush the flooded frame + 1008 close
+        frames: list[dict] = []
+        with pytest.raises(WebSocketDisconnect) as excinfo:
+            while True:
+                frames.append(_recv_status(ws))
+    assert excinfo.value.code == 1008
+    assert frames == [{"type": "flooded", "reason": "cap_exceeded", "count": 6}]
+
+    # Trace is durably intact up to the cap — no silent drop corrupted it.
+    assert [e.seq for e in store.get_trace(LEARNER, TASK)] == [0, 1, 2, 3, 4]
+
+    # The integrity signal Proctor/Phase-3 grader consume (G-4).
+    assert app.state.trace_integrity.is_incomplete(LEARNER, TASK) is True
+    assert app.state.trace_integrity.reason(LEARNER, TASK) == "INCOMPLETE_FLOODED"
+
+    # ...and it is readable over HTTP (trace with events + flag → 200, not 404).
+    body = client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}").json()
+    assert body["incomplete"] is True
+    assert body["integrity_reason"] == "INCOMPLETE_FLOODED"
+
+
+def test_queue_overflow_also_floods(
+    client: TestClient, app, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """G-3: inbound-queue overflow → same 1008 + INCOMPLETE_FLOODED (never
+    drop-oldest). Lower the queue bound so one burst trips it without needing
+    256 frames."""
+    monkeypatch.setattr(ingest_mod, "INBOUND_QUEUE_MAX", 1)
+    with client.websocket_connect(_ingest_url(learner_id="L-q", task_id="T-q")) as ws:
+        for seq in range(64):  # receiver out-drains the drainer with bound=1
+            ws.send_text(_frame(seq))
+        _drain(ws)
+        with pytest.raises(WebSocketDisconnect) as excinfo:
+            while True:
+                _recv_status(ws)
+    assert excinfo.value.code == 1008
+    assert app.state.trace_integrity.is_incomplete("L-q", "T-q") is True
+    assert app.state.trace_integrity.reason("L-q", "T-q") == "INCOMPLETE_FLOODED"
+
+
+@pytest.mark.asyncio
+async def test_queue_overflow_flood_session_task_terminates(tmp_path, monkeypatch):
+    """P7 regression: the queue-overflow flood path must not LEAK the
+    session coroutine. v0.3's receiver returned from its QueueFull branch
+    without the disconnect sentinel, so the drainer parked on an empty queue
+    forever and IngestSession.run() never returned — one leaked
+    (pinger+drainer) task-set per flooded trace, unbounded over a long-lived
+    process. A REAL uvicorn server (TestClient teardown hides the leak) is
+    stopped after the flood; the session tasks must be gone shortly after.
+    """
+    import asyncio
+    import contextlib
+    import socket as socket_mod
+
+    import uvicorn
+
+    from ai_service.main import create_app
+    from ai_service.telemetry.ingest import TraceIntegrityMap
+    from ai_service.telemetry.store import SQLiteTraceStore
+
+    monkeypatch.setattr(ingest_mod, "INBOUND_QUEUE_MAX", 1)
+    store = SQLiteTraceStore(db_path=tmp_path / "leak.db")
+    app = create_app(
+        Settings(
+            provider="mock",
+            db_path=tmp_path / "leak.db",
+            sandbox_dir=tmp_path / "sandboxes",
+        )
+    )
+    app.state.trace_store = store
+    app.state.trace_integrity = TraceIntegrityMap()
+
+    with socket_mod.socket() as s:
+        s.bind(("127.0.0.1", 0))
+        port = s.getsockname()[1]
+    server = uvicorn.Server(
+        uvicorn.Config(app, host="127.0.0.1", port=port, log_level="warning")
+    )
+    serve_task = asyncio.get_running_loop().create_task(server.serve())
+    leaked = True
+    try:
+        for _ in range(100):
+            if server.started:
+                break
+            await asyncio.sleep(0.1)
+        assert server.started
+
+        import websockets
+
+        uri = (
+            f"ws://127.0.0.1:{port}/v1/telemetry/ingest"
+            f"?learner_id=L-leak&task_id=T-leak"
+        )
+        async with websockets.connect(uri) as ws:
+            for seq in range(64):  # bound=1 → guaranteed overflow
+                await ws.send(_frame(seq, sandbox_id=""))
+            # The flood close (1008) reaches the client.
+            try:
+                await asyncio.wait_for(ws.recv(), timeout=10.0)
+                await asyncio.wait_for(ws.recv(), timeout=10.0)
+            except (websockets.exceptions.ConnectionClosed, TimeoutError, OSError):
+                pass
+        assert app.state.trace_integrity.is_incomplete("L-leak", "T-leak")
+
+        # The session's run() must have returned: no lingering nc-* tasks
+        # holding the socket open. Poll briefly — teardown is async.
+        deadline = asyncio.get_running_loop().time() + 5.0
+        while asyncio.get_running_loop().time() < deadline:
+            names = {
+                t.get_name()
+                for t in asyncio.all_tasks()
+                if t is not asyncio.current_task()
+            }
+            if not any("ingest" in n.lower() for n in names):
+                leaked = False
+                break
+            await asyncio.sleep(0.1)
+    finally:
+        server.should_exit = True
+        with contextlib.suppress(Exception):
+            await asyncio.wait_for(serve_task, timeout=10.0)
+        store.close()
+    assert not leaked, "IngestSession task set leaked after queue-overflow flood"
+
+
+def test_reconnect_after_flood_cannot_resurrect_trace(
+    client: TestClient, app, store: SQLiteTraceStore
+) -> None:
+    """INCOMPLETE_FLOODED is terminal: a fresh connection for the same trace
+    is immediately closed 1008 — intake never resumes on a flagged trace."""
+    with client.websocket_connect(_ingest_url()) as ws:
+        for seq in range(6):
+            ws.send_text(_frame(seq))
+        _drain(ws)
+        with pytest.raises(WebSocketDisconnect):
+            while True:
+                _recv_status(ws)
+    assert store.latest_seq(LEARNER, TASK) == 4
+
+    with client.websocket_connect(_ingest_url()) as ws:
+        ws.send_text(_frame(5))
+        _drain(ws)
+        with pytest.raises(WebSocketDisconnect) as excinfo:
+            while True:
+                _recv_status(ws)
+    assert excinfo.value.code == 1008
+    assert store.latest_seq(LEARNER, TASK) == 4  # nothing appended post-flag
+
+
+# -- frame validation ------------------------------------------------------------
+
+
+def test_frame_with_learner_id_in_body_rejected(client: TestClient) -> None:
+    """Identity lives in the URL (extra=forbid): a frame carrying learner_id
+    is a contract violation → event_rejected, nothing stored."""
+    with client.websocket_connect(_ingest_url()) as ws:
+        bad = json.loads(_frame(0))
+        bad["learner_id"] = "spoofed"
+        ws.send_text(json.dumps(bad))
+        _drain(ws)
+        frames = [_recv_status(ws)]
+    assert frames[0]["type"] == "event_rejected"
+    assert client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}").status_code == 404
+
+
+def test_unknown_event_kind_rejected_nothing_stored(client: TestClient) -> None:
+    with client.websocket_connect(_ingest_url()) as ws:
+        ws.send_text(_frame(0, kind="rm_rf_everything"))
+        _drain(ws)
+        frames = [_recv_status(ws)]
+    assert frames[0]["type"] == "event_rejected"
+    assert "unknown event kind" in frames[0]["detail"]
+    assert client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}").status_code == 404
+
+
+# -- unknown trace reads ---------------------------------------------------------
+
+
+def test_unknown_trace_get_returns_404(client: TestClient) -> None:
+    assert client.get("/v1/telemetry/traces/nobody/never").status_code == 404
+    assert client.get("/v1/telemetry/gaps/nobody/never").status_code == 404
+
+
+# -- handshake contract ------------------------------------------------------------
+
+
+def test_missing_identity_query_params_rejected_at_handshake(
+    client: TestClient,
+) -> None:
+    with pytest.raises(WebSocketDisconnect) as excinfo:
+        with client.websocket_connect("/v1/telemetry/ingest"):
+            pass
+    assert excinfo.value.code == 1008
+
+
+def test_browser_origin_rejected_in_explicit_list_mode(
+    settings, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """D-038: with an explicit AI_CORS_ORIGINS list, a page loaded in the
+    learner's browser (unlisted Origin) must not be able to open the ingest
+    socket and poison/flood the trace. The stdlib capture agent sends no
+    Origin and is unaffected (see the no-origin test below)."""
+    from fastapi.testclient import TestClient as TC
+
+    restricted = settings.model_copy(update={"cors_origins": "http://localhost:3000"})
+    app = create_app(restricted)
+    with TC(app) as c:
+        with pytest.raises(WebSocketDisconnect) as excinfo:
+            with c.websocket_connect(
+                _ingest_url(), headers={"Origin": "https://evil.example"}
+            ):
+                pass
+        assert excinfo.value.code == 1008
+
+
+def test_wildcard_mode_admits_any_browser_origin(client: TestClient) -> None:
+    """D-038 default ('*'): remote-browser origins open the ingest socket —
+    the remote build surface streams telemetry from the learner's browser."""
+    with client.websocket_connect(
+        _ingest_url(), headers={"Origin": "http://nextcraft-1:3000"}
+    ) as ws:
+        ws.send_text(_frame(0))
+
+
+def test_dev_origin_and_no_origin_both_allowed(client: TestClient) -> None:
+    """The same-origin dev page (Next.js :3000) opens fine, and so does the
+    capture-agent path (no Origin header at all)."""
+    for headers in ({"Origin": "http://localhost:3000"}, {}):
+        with client.websocket_connect(_ingest_url(), headers=headers) as ws:
+            ws.send_text(_frame(0))
+        body = client.get(f"/v1/telemetry/traces/{LEARNER}/{TASK}").json()
+        assert [e["seq"] for e in body["events"]] == [0]
+        # Unique trace per iteration would collide on (LEARNER, TASK) PK —
+        # seq 0 re-sent is deduped, so one row is the invariant either way.
+        assert len(body["events"]) == 1
+
+
+# -- keepalive ---------------------------------------------------------------------
+
+
+def test_server_sends_periodic_ping(
+    client: TestClient, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """Keepalive (D-026): the server emits a ping on PING_INTERVAL_S so the
+    stdlib capture agent's auto-pong keeps the path warm. Frame content is
+    opaque to the agent; the contract is that a ping arrives while idle."""
+    monkeypatch.setattr(ingest_mod, "PING_INTERVAL_S", 0.05)
+    with client.websocket_connect(_ingest_url()) as ws:
+        ping = ws.receive_bytes()  # first server→client frame while idle
+    assert ping  # opaque payload; contract is "a ping frame arrives"

apps/ai-service/tests/api/test_variants.py

@@ -0,0 +1,277 @@
+"""Variant API tests — generation, cache, distinctness over HTTP (Task 4-3-01).
+
+Contract under test (api/variants.py, REQ-3-005):
+
+    POST /v1/variants {learner_id, template_id}
+        first request       → 200 generated variant (task_id, seed, params,
+                              statement, starter_files, competency_id)
+        repeat request      → 200 the SAME cached variant with ZERO LLM
+                              calls (D-029 reproducibility through the
+                              whole HTTP stack)
+        unknown template    → 404
+    POST /v1/variants {learner_id, competency_id}
+        bound competency    → 200 the first template for that competency
+        unknown competency  → 404
+        neither id given    → 422
+    GET  /v1/variants/{task_id}
+        generated earlier  → 200 the stored variant (generate→get roundtrip)
+        unknown task       → 404
+    GET  /v1/variants?learner_id=...
+        → 200 that learner's variants only (scoping); [] for a learner
+          with none.
+
+Distinctness (REQ-3-005, API level): two learners POSTing the same
+template receive distinct statements, seeds and task_ids, and GET by
+task_id hands each back their own variant.
+
+Wiring: per-test tmp-path SQLiteVariantStore + a pre-set VariantGenerator
+(state-injection override — the lifespan adopts variant_store /
+variant_generator from app.state instead of constructing them; same
+pattern as test_grading.py / test_telemetry_ingest.py). The generator
+binds a ScriptedRenderProvider so each test controls the render LLM
+exactly, and counts calls so the cache test can assert the LLM was never
+reached on the second POST.
+
+Zero network: providers are MockProvider family members only (conftest
+rule, enforced in _make_client).
+"""
+
+from __future__ import annotations
+
+from collections.abc import Iterator
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.config import Settings
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.variants.generator import VariantGenerator
+from ai_service.variants.store import SQLiteVariantStore
+from ai_service.variants.templates import get_template
+
+LEARNER_A = "variant-learner-a"
+LEARNER_B = "variant-learner-b"
+TEMPLATE = "tpl-llm-judge"
+COMPETENCY = "stack-orchestration-c007"  # tpl-llm-judge's D-021 binding
+
+
+class ScriptedRenderProvider(MockProvider):
+    """Deterministic render whose statement embeds the params (distinct per
+    draw); counts calls so the cache test asserts zero LLM calls on the
+    second POST. Mirrors tests/variants/test_generator.py."""
+
+    def __init__(self) -> None:
+        super().__init__()
+        self.calls = 0
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):  # noqa: ANN001
+        self.calls += 1
+        import json
+
+        user = next(m.content for m in reversed(messages) if m.role == "user")
+        # Distinct per distinct params: hash the seeded slot lines.
+        fingerprint = abs(hash(user)) % 10_000
+        return json.dumps({"statement": f"Scripted variant #{fingerprint} — build it."})
+
+
+@pytest.fixture()
+def store(tmp_path: Path) -> Iterator[SQLiteVariantStore]:
+    s = SQLiteVariantStore(db_path=tmp_path / "variants.db")
+    yield s
+    s.close()
+
+
+@pytest.fixture()
+def provider() -> ScriptedRenderProvider:
+    return ScriptedRenderProvider()
+
+
+def _make_client(
+    tmp_path: Path, store: SQLiteVariantStore, provider: MockProvider
+) -> TestClient:
+    """App + TestClient with a pre-set store + generator.
+
+    The lifespan adopts both (state-injection override); the generator
+    binds OUR provider, so the fixture — not Settings — scripts the LLM.
+    Cloud-free guard: the provider must be a MockProvider family member
+    (conftest rule, enforced here because this module builds its own
+    client rather than consuming the conftest one).
+    """
+    assert isinstance(provider, MockProvider)
+    settings = Settings(
+        provider="mock",
+        db_path=tmp_path / "variant-test.db",
+        sandbox_dir=tmp_path / "sandboxes",
+    )
+    app = create_app(settings)
+    app.state.variant_store = store
+    if getattr(app.state, "variant_generator", None) is None:
+        app.state.variant_generator = VariantGenerator(
+            store, provider, model="gemma4:31b"
+        )
+    return TestClient(app)
+
+
+def _post(
+    client: TestClient,
+    learner_id: str,
+    template_id: str | None = TEMPLATE,
+    competency_id: str | None = None,
+):
+    return client.post(
+        "/v1/variants",
+        json={
+            "learner_id": learner_id,
+            **({"template_id": template_id} if template_id is not None else {}),
+            **({"competency_id": competency_id} if competency_id is not None else {}),
+        },
+    )
+
+
+# -- POST: generation + response shape --------------------------------------------
+
+
+class TestGenerate:
+    def test_post_returns_full_variant_payload(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = _post(client, LEARNER_A)
+
+        assert response.status_code == 200
+        body = response.json()
+        assert body["learner_id"] == LEARNER_A
+        assert body["template_id"] == TEMPLATE
+        assert body["competency_id"] == COMPETENCY  # enrichment from get_template
+        assert body["task_id"].startswith("task-") and len(body["task_id"]) == len("task-") + 16
+        assert body["seed"]  # non-empty D-029 seed
+        assert body["statement"]  # rendered, non-empty
+        assert body["starter_files"] == get_template(TEMPLATE).starter_files
+        assert body["params"]  # seeded slot values
+        assert body["created_at"]  # ISO timestamp travels
+        assert provider.calls == 1  # first POST renders exactly once
+
+    def test_post_then_get_roundtrip(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            posted = _post(client, LEARNER_A)
+            assert posted.status_code == 200
+
+            fetched = client.get(f"/v1/variants/{posted.json()['task_id']}")
+
+        assert fetched.status_code == 200
+        assert fetched.json() == posted.json()  # identical stored variant
+
+    def test_post_regenerate_is_cache_hit_no_llm_call(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            first = _post(client, LEARNER_A)
+            assert first.status_code == 200
+            assert provider.calls == 1
+
+            second = _post(client, LEARNER_A)  # same (learner, template) → cache
+            assert second.status_code == 200
+            assert second.json() == first.json()  # D-029: the SAME variant
+            assert provider.calls == 1, "cached regenerate must not render again"
+
+
+# -- POST: distinctness (REQ-3-005, API level) ------------------------------------
+
+
+class TestDistinctLearners:
+    def test_two_learners_distinct_statements_and_task_ids(
+        self, tmp_path, store, provider
+    ):
+        with _make_client(tmp_path, store, provider) as client:
+            a = _post(client, LEARNER_A)
+            b = _post(client, LEARNER_B)
+            assert a.status_code == 200 and b.status_code == 200
+
+            a_body, b_body = a.json(), b.json()
+
+            # ...and each learner GETs back exactly their own variant
+            a_fetched = client.get(f"/v1/variants/{a_body['task_id']}")
+            b_fetched = client.get(f"/v1/variants/{b_body['task_id']}")
+
+        assert a_body["statement"] != b_body["statement"]
+        assert a_body["seed"] != b_body["seed"]
+        assert a_body["task_id"] != b_body["task_id"]
+        assert a_body["competency_id"] == b_body["competency_id"]  # same bar (a-5)
+        assert a_fetched.status_code == 200 and a_fetched.json() == a_body
+        assert b_fetched.status_code == 200 and b_fetched.json() == b_body
+
+
+# -- POST: template / competency resolution ---------------------------------------
+
+
+class TestTemplateResolution:
+    def test_unknown_template_404(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = _post(client, LEARNER_A, template_id="tpl-does-not-exist")
+
+        assert response.status_code == 404
+        assert "no task template" in response.json()["detail"]
+
+    def test_competency_lookup_uses_first_bound_template(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = _post(
+                client, LEARNER_A, template_id=None, competency_id=COMPETENCY
+            )
+
+        assert response.status_code == 200
+        body = response.json()
+        assert body["template_id"] == TEMPLATE  # the competency's first template
+        assert body["competency_id"] == COMPETENCY
+
+    def test_unknown_competency_404(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = _post(
+                client, LEARNER_A, template_id=None, competency_id="stack-none-c999"
+            )
+
+        assert response.status_code == 404
+        assert "no task template for competency" in response.json()["detail"]
+
+    def test_neither_id_given_422(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = _post(client, LEARNER_A, template_id=None, competency_id=None)
+
+        assert response.status_code == 422
+
+
+# -- GET: stored reads ------------------------------------------------------------
+
+
+class TestStoredReads:
+    def test_get_unknown_task_404(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = client.get("/v1/variants/task-0000000000000000")
+
+        assert response.status_code == 404
+        assert "no stored variant" in response.json()["detail"]
+
+    def test_list_scoped_to_one_learner(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            assert _post(client, LEARNER_A).status_code == 200
+            assert _post(
+                client, LEARNER_A, template_id="tpl-guardrail-schema"
+            ).status_code == 200
+            assert _post(client, LEARNER_B).status_code == 200
+
+            listed = client.get("/v1/variants", params={"learner_id": LEARNER_A})
+            empty = client.get("/v1/variants", params={"learner_id": "nobody"})
+
+        assert listed.status_code == 200
+        variants = listed.json()["variants"]
+        assert len(variants) == 2  # LEARNER_A's two, LEARNER_B's excluded
+        assert {v["learner_id"] for v in variants} == {LEARNER_A}
+        assert {v["template_id"] for v in variants} == {TEMPLATE, "tpl-guardrail-schema"}
+        # chronological ordering; both carry the competency enrichment
+        assert all(v["competency_id"] for v in variants)
+
+        assert empty.status_code == 200
+        assert empty.json()["variants"] == []
+
+    def test_list_missing_learner_param_422(self, tmp_path, store, provider):
+        with _make_client(tmp_path, store, provider) as client:
+            response = client.get("/v1/variants")
+
+        assert response.status_code == 422

apps/ai-service/tests/grading/__init__.py

@@ -0,0 +1 @@
+"""Process-trace grading tests (REQ-3-004)."""

apps/ai-service/tests/grading/test_calibration.py

@@ -0,0 +1,361 @@
+"""Grading calibration CONTRACT test (Task 3-2-02, REQ-3-004, D-021).
+
+HONEST SCOPE — what this test is and is NOT:
+
+This is a CALIBRATION CONTRACT test, not an LLM quality test. The mock
+provider scripts archetype-mapped rubric scores for each fixture; the
+assertions verify that:
+
+  1. the ENGINE pipeline (gate → digest → prompt → D-020 → validate →
+     persist) runs each archetype end-to-end and preserves the score
+     ordering the scripts impose;
+  2. the fixtures' digest FEATURES actually separate the archetypes
+     deterministically (the non-LLM half of the calibration: paste-and-run
+     vs iterative debugging vs never-passing MUST produce observably
+     different digests — otherwise no rubric could ever separate them);
+  3. fixture IDs are valid corpus-aligned strings (D-021).
+
+It does NOT prove the real LLM scores archetypes in this order — that
+requires live-model evaluation, which is out of scope for v0.3 tests
+(tests NEVER call the cloud; conftest enforces). What the contract pins
+is the ORDERING the rubric + digest must eventually enforce:
+
+    strong.process_quality >= lazy.process_quality
+    strong.correctness     >  struggling.correctness
+
+If a future prompt/digest change makes the scripted mapping unachievable
+by ANY model (e.g. the digest stops separating the archetypes), test 2
+fails — that is the calibration signal.
+"""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+import pytest
+
+from ai_service.corpus.telemetry import LAB_SCENARIOS
+from ai_service.corpus.trace_fixtures import (
+    CALIBRATION_LEARNER_ID,
+    LAZY_BUILDER,
+    STRONG_BUILDER,
+    STRUGGLING_BUILDER,
+    TRACE_FIXTURES,
+    digest_of,
+)
+from ai_service.grading.engine import GradingEngine
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+
+#: Archetype-mapped rubric scripts (the ORDERING CONTRACT in numbers).
+#: strong: iterative verified building — highest process quality.
+#: lazy: ends pass with zero iteration — the a-4 negative (churn with no
+#:        test progress caps process_quality at 1; here it's paste-churn +
+#:        one late test, so 1). correctness lands high (final pass, fast).
+#: struggling: never passes — lowest correctness.
+ARCHETYPE_SCORES: dict[str, dict] = {
+    "strong": {
+        "criteria": {
+            "process_quality": 4,
+            "correctness": 4,
+            "debugging_discipline": 4,
+            "test_usage": 4,
+        },
+        "strengths": ["tight edit-test loops; every failure cycle closed"],
+        "gaps": ["session ended without a final full-suite run"],
+        "verdict": "mastered",
+    },
+    "lazy": {
+        "criteria": {
+            "process_quality": 1,  # a-4: bulk paste + churn, one late test
+            "correctness": 3,
+            "debugging_discipline": 1,
+            "test_usage": 1,
+        },
+        "strengths": ["final tests pass on the first run"],
+        "gaps": ["one bulk paste with zero verified increments"],
+        "verdict": "developing",
+    },
+    "struggling": {
+        "criteria": {
+            "process_quality": 2,
+            "correctness": 0,  # never passes
+            "debugging_discipline": 1,  # cycles never close
+            "test_usage": 3,
+        },
+        "strengths": ["persisted through repeated failures"],
+        "gaps": ["the same failure recurred three times"],
+        "verdict": "not_yet",
+    },
+}
+
+
+class ArchetypeProvider:
+    """Mock provider scripting archetype-mapped scores keyed by fixture.
+
+    Keying on the digest embedded in the user turn (not on call order)
+    keeps the mapping robust to gate-path noise and asserts the prompt
+    really carries the fixture's digest. Records the last user turn so
+    tests can re-assert D-028 (digest in prompt; raw events out).
+    """
+
+    def __init__(self) -> None:
+        self.digest_to_archetype: dict[str, str] = {}
+        self.calls = 0
+        self.last_user_prompt: str = ""
+
+    def register(self, fixture) -> None:
+        self.digest_to_archetype[digest_of(fixture).model_dump_json()] = (
+            fixture.archetype
+        )
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):
+        self.calls += 1
+        user = next((m for m in reversed(messages) if m.role == "user"), None)
+        assert user is not None
+        self.last_user_prompt = user.content
+        archetype = None
+        for digest_json, arch in self.digest_to_archetype.items():
+            if digest_json in user.content:
+                archetype = arch
+                break
+        assert archetype is not None, "fixture digest not found in prompt"
+        return json.dumps(ARCHETYPE_SCORES[archetype])
+
+    async def stream_chat(self, messages, *, model, temperature=0.7, response_format=None):
+        yield await self.chat(
+            messages, model=model, temperature=temperature, response_format=response_format
+        )
+
+
+@pytest.fixture
+def trace_store(tmp_path: Path) -> SQLiteTraceStore:
+    store = SQLiteTraceStore(db_path=tmp_path / "traces.db")
+    yield store
+    store.close()
+
+
+@pytest.fixture
+def grade_store(tmp_path: Path) -> SQLiteGradeStore:
+    store = SQLiteGradeStore(db_path=tmp_path / "grades.db")
+    yield store
+    store.close()
+
+
+@pytest.fixture
+def provider() -> ArchetypeProvider:
+    p = ArchetypeProvider()
+    for fixture in (STRONG_BUILDER, LAZY_BUILDER, STRUGGLING_BUILDER):
+        p.register(fixture)
+    return p
+
+
+@pytest.fixture
+def engine(trace_store, grade_store, provider) -> GradingEngine:
+    return GradingEngine(
+        trace_store, grade_store, TraceIntegrityMap(), provider, model="gemma4:31b"
+    )
+
+
+async def _grade_fixture(engine, trace_store, fixture) -> dict:
+    for event in fixture.events:
+        trace_store.append(event)
+    record = await engine.grade(fixture.learner_id, fixture.task_id)
+    assert record.verdict == "GRADED", (
+        f"fixture {fixture.fixture_id} did not grade cleanly: {record.verdict}"
+    )
+    return record.scores
+
+
+class TestFixtureAlignment:
+    """D-021: fixture IDs align with the v0.2 corpus scenario IDs."""
+
+    def test_every_fixture_id_is_corpus_scenario_scoped(self):
+        for fixture in TRACE_FIXTURES.values():
+            assert fixture.fixture_id.count("::") == 1
+            scenario_part, trace_part = fixture.fixture_id.split("::")
+            assert scenario_part in LAB_SCENARIOS, (
+                f"{scenario_part!r} is not a v0.2 lab scenario id"
+            )
+            assert trace_part.endswith("-trace")
+            # double-brace-free single slug
+            assert "-" in trace_part
+
+    def test_every_fixture_aligns_to_an_existing_scenario(self):
+        assert set(f.aligned_scenario_id for f in TRACE_FIXTURES.values()) <= set(
+            LAB_SCENARIOS
+        )
+
+    def test_fixture_ids_are_valid_strings(self):
+        for fixture in TRACE_FIXTURES.values():
+            assert fixture.fixture_id and isinstance(fixture.fixture_id, str)
+            assert fixture.learner_id == CALIBRATION_LEARNER_ID
+            assert fixture.learner_id.startswith("learner-00")
+
+    def test_lazy_alignment_is_the_flagged_scenario(self):
+        # v0.2's paste-and-run archetype is the flagged lab scenario (D-021).
+        assert LAZY_BUILDER.aligned_scenario_id == "lab-scenario-flagged"
+        assert LAZY_BUILDER.archetype == "lazy"
+
+    def test_all_three_archetypes_present(self):
+        assert {f.archetype for f in TRACE_FIXTURES.values()} == {
+            "strong",
+            "lazy",
+            "struggling",
+        }
+
+    def test_events_are_valid_telemetry_events(self):
+        from ai_service.telemetry.models import TelemetryEvent
+
+        for fixture in TRACE_FIXTURES.values():
+            assert len(fixture.events) > 0
+            for e in fixture.events:
+                assert isinstance(e, TelemetryEvent)
+                assert e.seq >= 0
+            # contiguous seqs: a grading trace must clear the G-4 gate
+            seqs = [e.seq for e in fixture.events]
+            assert seqs == list(range(len(seqs))), (
+                f"{fixture.fixture_id} has seq gaps — it would trip the gate"
+            )
+
+
+class TestDigestSeparatesArchetypes:
+    """The non-LLM half of calibration: digests differ in the RIGHT direction.
+
+    If these deterministic assertions ever fail, no rubric — however good —
+    could separate the archetypes: the digest would be lossy in the wrong
+    places. These are the features the a-4 advisory and the rubric anchors
+    speak about, computed over REAL trace fixtures.
+    """
+
+    def test_strong_shows_iterative_cycles_lazy_shows_none(self):
+        strong = digest_of(STRONG_BUILDER)
+        lazy = digest_of(LAZY_BUILDER)
+        assert strong.error_fix_cycles >= 1
+        assert lazy.error_fix_cycles == 0
+
+    def test_strong_tests_early_lazy_tests_late(self):
+        strong = digest_of(STRONG_BUILDER)
+        lazy = digest_of(LAZY_BUILDER)
+        assert strong.first_test_pass_offset_s is not None
+        assert lazy.first_test_pass_offset_s is not None
+        assert strong.first_test_pass_offset_s < lazy.first_test_pass_offset_s
+
+    def test_struggling_never_passes(self):
+        struggling = digest_of(STRUGGLING_BUILDER)
+        assert struggling.final_test_status == "fail"
+        assert struggling.test_pass_count == 0
+        assert struggling.test_fail_count >= 3
+
+    def test_strong_and_struggling_both_end(self):
+        strong = digest_of(STRONG_BUILDER)
+        assert strong.final_test_status == "pass"
+        assert strong.test_fail_count >= 1  # honest failure, then closed
+
+    def test_lazy_churns_without_test_progress(self):
+        """a-4 anchor: bulk edit + command churn with (almost) no test signal."""
+        lazy = digest_of(LAZY_BUILDER)
+        assert lazy.edit_count >= 2
+        assert lazy.test_fail_count == 0
+        assert lazy.command_count >= 2
+        assert lazy.final_test_status == "pass"  # ...and still one late test
+
+
+class TestOrderingContract:
+    """The scripted ORDERING CONTRACT through the full engine pipeline."""
+
+    async def test_strong_at_least_lazy_on_process_quality(
+        self, engine, trace_store
+    ):
+        strong = await _grade_fixture(engine, trace_store, STRONG_BUILDER)
+        lazy = await _grade_fixture(engine, trace_store, LAZY_BUILDER)
+        assert (
+            strong["criteria"]["process_quality"]
+            >= lazy["criteria"]["process_quality"]
+        )
+
+    async def test_strong_above_struggling_on_correctness(self, engine, trace_store):
+        strong = await _grade_fixture(engine, trace_store, STRONG_BUILDER)
+        struggling = await _grade_fixture(engine, trace_store, STRUGGLING_BUILDER)
+        assert strong["criteria"]["correctness"] > struggling["criteria"]["correctness"]
+
+    async def test_full_ordering_across_all_three_archetypes(
+        self, engine, trace_store, grade_store
+    ):
+        scores = {
+            arch: await _grade_fixture(engine, trace_store, f)
+            for arch, f in (
+                ("strong", STRONG_BUILDER),
+                ("lazy", LAZY_BUILDER),
+                ("struggling", STRUGGLING_BUILDER),
+            )
+        }
+        # the required ordering contract (task 3-2-02):
+        assert (
+            scores["strong"]["criteria"]["process_quality"]
+            >= scores["lazy"]["criteria"]["process_quality"]
+        )
+        assert (
+            scores["strong"]["criteria"]["correctness"]
+            > scores["struggling"]["criteria"]["correctness"]
+        )
+        # Deliberate NON-ordering, asserted to document the calibration
+        # intent: struggling > lazy on process quality. Process quality is
+        # not outcome — the struggling builder iterated (edits between
+        # runs, three test runs), the lazy builder only pasted once and
+        # verified once (a-4 caps that at 1). A rubric that scores a
+        # never-passing iterator BELOW a paste-and-runner on process is
+        # miscalibrated on its own anchors.
+        assert (
+            scores["struggling"]["criteria"]["process_quality"]
+            > scores["lazy"]["criteria"]["process_quality"]
+        )
+        # verdicts also ordered: mastered >= developing > not_yet
+        assert scores["strong"]["verdict"] == "mastered"
+        assert scores["lazy"]["verdict"] == "developing"
+        assert scores["struggling"]["verdict"] == "not_yet"
+
+        # all three persisted under the calibration learner, latest-state
+        grades = grade_store.list_for_learner(CALIBRATION_LEARNER_ID)
+        assert len(grades) == 3
+        assert {g.task_id for g in grades} == {
+            "task-calibration-strong",
+            "task-calibration-lazy",
+            "task-calibration-struggling",
+        }
+
+    async def test_prompt_carries_fixture_digest_not_raw_events(
+        self, engine, trace_store, provider
+    ):
+        """D-028 on the calibration path: digest in the prompt, raw events
+        never — the lazy fixture plants distinctive payload strings
+        ("eval.py", the bulk 240-line paste) that must NOT appear."""
+        for event in LAZY_BUILDER.events:
+            trace_store.append(event)
+        await engine.grade(LAZY_BUILDER.learner_id, LAZY_BUILDER.task_id)
+
+        assert provider.calls == 1
+        prompt = provider.last_user_prompt
+        # the fixture's digest reached the LLM...
+        assert "PROCESS TRACE DIGEST (JSON):" in prompt
+        assert '"error_fix_cycles"' in prompt
+        # ...but no raw trace material or identity did (D-028)
+        assert "eval.py" not in prompt
+        assert LAZY_BUILDER.learner_id not in prompt
+        assert LAZY_BUILDER.task_id not in prompt
+
+    async def test_struggling_digests_reach_engine_without_gate(
+        self, engine, trace_store, provider
+    ):
+        for event in STRUGGLING_BUILDER.events:
+            trace_store.append(event)
+        record = await engine.grade(
+            STRUGGLING_BUILDER.learner_id, STRUGGLING_BUILDER.task_id
+        )
+        assert record.verdict == "GRADED"  # honest struggle is gradable
+        assert record.scores["criteria"]["correctness"] == 0
+        # a never-passing session still yields a real verdict — the grade
+        # is low, not missing
+        assert record.scores["verdict"] == "not_yet"

apps/ai-service/tests/grading/test_engine.py

@@ -0,0 +1,528 @@
+"""GradingEngine tests (Task 3-2-01, REQ-3-004, G-4 binding).
+
+Contract under test:
+    - G-4 gate FIRST: seq gaps OR integrity-flagged trace →
+      verdict=UNGRADABLE_TRACE_INCOMPLETE with the gap list surfaced in
+      scores; the LLM is NEVER called on those paths (asserted). Empty
+      trace → UNGRADABLE_EMPTY_TRACE. All are first-class GradeRecords
+      (returned + persisted), not exceptions.
+    - Complete trace: prompt carries the digest JSON but NO raw trace
+      material (D-028) — a distinctive marker planted in a command payload
+      must be absent from every message the provider received.
+    - Malformed LLM JSON → the D-020 retry path recovers (scripted bad
+      first, good second) → graded.
+    - Re-grade overwrites the stored record (GradeStore upsert).
+
+Zero network: everything runs against the mock provider (conftest rule).
+"""
+
+from __future__ import annotations
+
+import json
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+import pytest
+from pydantic import BaseModel, ValidationError
+
+from ai_service.grading.engine import GradingEngine, RubricScore
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.types import Message
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.models import TelemetryEvent
+from ai_service.telemetry.store import SQLiteTraceStore
+from ai_service.variants.store import SQLiteVariantStore
+
+T0 = datetime(2026, 9, 12, 1, 0, 0, tzinfo=UTC)
+RAW_MARKER = "SECRET-COMMAND-MARKER-7f3a"
+
+
+class RubricJSON:
+    """Canonical well-formed rubric payload (matches engine.RubricScore)."""
+
+    PAYLOAD: dict = {
+        "criteria": {
+            "process_quality": 3,
+            "correctness": 4,
+            "debugging_discipline": 3,
+            "test_usage": 4,
+        },
+        "strengths": ["tight edit-test loops throughout"],
+        "gaps": ["final commit discipline loose"],
+        "verdict": "mastered",
+    }
+
+
+class RecordingProvider:
+    """Mock LLMProvider: scripted replies + captured requests (no network).
+
+    Subclass-composes ai_service.llm.mock.MockProvider so the conftest
+    cloud-free guard philosophy holds — but records every message list
+    and counts calls, which the stock mocks don't do.
+    """
+
+    def __init__(self) -> None:
+        from ai_service.llm.mock import MockProvider
+
+        self._mock = MockProvider()
+        self.replies: list[str] = []  # consumed left-to-right; last repeats
+        self.requests: list[list[Message]] = []
+        self.calls = 0
+
+    async def chat(self, messages, *, model, temperature=0.7, response_format=None):
+        self.calls += 1
+        self.requests.append([Message(role=m.role, content=m.content) for m in messages])
+        if self.replies:
+            reply = self.replies.pop(0)
+        else:
+            reply = json.dumps(RubricJSON.PAYLOAD)
+        return reply
+
+    async def stream_chat(self, messages, *, model, temperature=0.7, response_format=None):
+        yield await self.chat(
+            messages, model=model, temperature=temperature, response_format=response_format
+        )
+
+
+def _event(
+    seq: int,
+    kind: str,
+    payload: dict | None = None,
+    offset_s: float = 0.0,
+    *,
+    learner: str = "engine-learner",
+    task: str = "engine-task",
+) -> TelemetryEvent:
+    return TelemetryEvent(
+        learner_id=learner,
+        task_id=task,
+        seq=seq,
+        kind=kind,
+        payload=payload or {},
+        ts=T0 + timedelta(seconds=offset_s),
+        sandbox_id="sbx-engine",
+    )
+
+
+def _complete_trace(marker_command: bool = True) -> list[TelemetryEvent]:
+    """A healthy iterative trace; seq 0..N contiguous (no gaps).
+
+    `marker_command` plants a distinctive string inside a command payload
+    — the D-028 assertion target: raw trace material must never reach the
+    provider, so the marker must not appear in any captured request.
+    """
+    if marker_command:
+        cmd_payload = {"cmd": f"echo {RAW_MARKER} && pytest -q"}
+    else:
+        cmd_payload = {"cmd": "pytest -q"}
+    return [
+        _event(0, "activity", {"state": "starting"}, 0.0),
+        _event(1, "file_diff", {"path": "a.py", "added": 12}, 10.0),
+        _event(2, "command", cmd_payload, 20.0),
+        _event(3, "test_result", {"passed": False, "exit_code": 1}, 25.0),
+        _event(4, "file_diff", {"path": "a.py", "added": 4, "removed": 2}, 40.0),
+        _event(5, "command", {"cmd": "pytest -q"}, 60.0),
+        _event(6, "test_result", {"passed": True, "exit_code": 0}, 65.0),
+    ]
+
+
+@pytest.fixture
+def trace_store(tmp_path: Path) -> SQLiteTraceStore:
+    store = SQLiteTraceStore(db_path=tmp_path / "traces.db")
+    yield store
+    store.close()
+
+
+@pytest.fixture
+def grade_store(tmp_path: Path) -> SQLiteGradeStore:
+    store = SQLiteGradeStore(db_path=tmp_path / "grades.db")
+    yield store
+    store.close()
+
+
+@pytest.fixture
+def integrity() -> TraceIntegrityMap:
+    return TraceIntegrityMap()
+
+
+@pytest.fixture
+def provider() -> RecordingProvider:
+    return RecordingProvider()
+
+
+@pytest.fixture
+def engine(trace_store, grade_store, integrity, provider) -> GradingEngine:
+    return GradingEngine(
+        trace_store, grade_store, integrity, provider, model="gemma4:31b"
+    )
+
+
+def _ingest(store: SQLiteTraceStore, events: list[TelemetryEvent]) -> None:
+    for e in events:
+        store.append(e)
+
+
+class TestRubricScoreModel:
+    """The D-020 schema contract — engine-side validation rules."""
+
+    def _rubric(self, **overrides) -> dict:
+        payload = json.loads(json.dumps(RubricJSON.PAYLOAD))  # deep copy
+        payload.update(overrides)
+        return payload
+
+    def test_well_formed_payload_validates(self):
+        score = RubricScore.model_validate(RubricJSON.PAYLOAD)
+        assert score.criteria["process_quality"] == 3
+        assert score.verdict == "mastered"
+
+    def test_unknown_criterion_rejected(self):
+        bad = self._rubric()
+        bad["criteria"]["extra_criterion"] = 2
+        with pytest.raises(ValidationError, match="criteria keys must be exactly"):
+            RubricScore.model_validate(bad)
+
+    def test_missing_criterion_rejected(self):
+        bad = self._rubric()
+        del bad["criteria"]["test_usage"]
+        with pytest.raises(ValidationError, match="criteria keys must be exactly"):
+            RubricScore.model_validate(bad)
+
+    def test_out_of_range_score_rejected(self):
+        bad = self._rubric()
+        bad["criteria"]["process_quality"] = 5
+        with pytest.raises(ValidationError, match="must be within 0-4"):
+            RubricScore.model_validate(bad)
+
+    def test_unknown_verdict_rejected(self):
+        bad = self._rubric()
+        bad["verdict"] = "excellent"
+        with pytest.raises(ValidationError, match="verdict must be one of"):
+            RubricScore.model_validate(bad)
+
+
+class TestCompleteTraceGrading:
+    async def test_graded_returns_validated_rubric_persisted_and_returned(
+        self, engine, trace_store, grade_store, provider
+    ):
+        _ingest(trace_store, _complete_trace())
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "GRADED"
+        assert record.model == "gemma4:31b"
+        assert record.scores == RubricJSON.PAYLOAD
+        assert record.scores["criteria"]["process_quality"] == 3
+        assert record.digest.get("error_fix_cycles") == 1
+        assert record.digest.get("final_test_status") == "pass"
+        # digest persisted for auditability (D-028 reproducible input)
+        assert record.digest.get("event_count") == 7
+
+        stored = grade_store.get("engine-learner", "engine-task")
+        assert stored is not None
+        assert stored.scores == record.scores
+        assert stored.verdict == "GRADED"
+
+    async def test_prompt_contains_digest_but_no_raw_trace(self, engine, trace_store, provider):
+        """D-028: the provider saw the digest JSON but NEVER the raw trace.
+
+        The marker was planted inside a command payload (seq 2); it must be
+        absent from every message of every request, while the digest marker
+        line + a JSON object with the digest's field names must be present.
+        """
+        _ingest(trace_store, _complete_trace(marker_command=True))
+
+        await engine.grade("engine-learner", "engine-task")
+
+        assert provider.calls == 1  # happy path: exactly one LLM call
+        all_messages = [m for req in provider.requests for m in req]
+        prompt_text = "\n".join(m.content for m in all_messages)
+
+        assert RAW_MARKER not in prompt_text, (
+            "raw trace material reached the LLM prompt (D-028 violation)"
+        )
+        # digest presence: marker line + a JSON object carrying digest fields
+        assert "PROCESS TRACE DIGEST (JSON):" in prompt_text
+        assert '"error_fix_cycles"' in prompt_text
+        assert '"final_test_status"' in prompt_text
+        # learner anonymity: identity strings never reach the LLM either
+        assert "engine-learner" not in prompt_text
+        assert "engine-task" not in prompt_text
+
+    async def test_system_prompt_has_rubric_and_a4_advisory(
+        self, engine, trace_store, provider
+    ):
+        _ingest(trace_store, _complete_trace())
+
+        await engine.grade("engine-learner", "engine-task")
+
+        system = provider.requests[0][0]
+        assert system.role == "system"
+        assert "process_quality" in system.content
+        assert "correctness" in system.content
+        assert "debugging_discipline" in system.content
+        assert "test_usage" in system.content
+        # a-4: churn-without-test-progress is a process-quality NEGATIVE
+        assert "churn is not" in system.content
+        # 0-4 level anchors present for each criterion
+        assert "ADVISORY" in system.content
+
+
+class TestGateFirst:
+    """G-4 binding: the gate fires BEFORE any grading work."""
+
+    async def test_gapped_trace_ungradable_llm_not_called(
+        self, engine, trace_store, provider, grade_store
+    ):
+        # seqs 0,1,3 stored — seq 2 missing (mirrors the task spec example)
+        events = _complete_trace()
+        gapped = [e for e in events if e.seq != 2]
+        _ingest(trace_store, gapped)
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "UNGRADABLE_TRACE_INCOMPLETE"
+        assert provider.calls == 0, "LLM was called despite a gapped trace (G-4)"
+        # gap list surfaced in scores
+        assert record.scores["missing_seqs"] == [2]
+        assert record.scores["integrity_flag"] is None
+        # first-class record: persisted, retrievable, empty digest
+        stored = grade_store.get("engine-learner", "engine-task")
+        assert stored is not None
+        assert stored.verdict == "UNGRADABLE_TRACE_INCOMPLETE"
+        assert stored.scores["missing_seqs"] == [2]
+        assert stored.digest == {}
+        assert stored.model == "none"
+
+    async def test_integrity_flagged_trace_ungradable_llm_not_called(
+        self, engine, trace_store, integrity, provider
+    ):
+        _ingest(trace_store, _complete_trace())
+        integrity.mark("engine-learner", "engine-task", "INCOMPLETE_FLOODED")
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "UNGRADABLE_TRACE_INCOMPLETE"
+        assert provider.calls == 0, "LLM was called despite integrity flag (G-4)"
+        assert record.scores["integrity_flag"] == "INCOMPLETE_FLOODED"
+        assert record.scores["missing_seqs"] == []  # complete rows, but untrusted
+
+    async def test_flag_wins_even_with_gaps(self, engine, trace_store, integrity, provider):
+        """Both gate signals set: the flag reason is surfaced (both listed)."""
+        events = _complete_trace()
+        _ingest(trace_store, [e for e in events if e.seq != 2])
+        integrity.mark("engine-learner", "engine-task", "INCOMPLETE_FLOODED")
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "UNGRADABLE_TRACE_INCOMPLETE"
+        assert provider.calls == 0
+        assert record.scores["integrity_flag"] == "INCOMPLETE_FLOODED"
+        assert record.scores["missing_seqs"] == [2]
+
+    async def test_empty_trace_ungradable(self, engine, provider, grade_store):
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "UNGRADABLE_EMPTY_TRACE"
+        assert provider.calls == 0, "LLM was called for an empty trace"
+        assert record.scores == {"integrity_flag": None, "missing_seqs": []}
+        assert record.digest == {}
+        stored = grade_store.get("engine-learner", "engine-task")
+        assert stored is not None
+        assert stored.verdict == "UNGRADABLE_EMPTY_TRACE"
+
+
+class TestD020RetryPath:
+    async def test_malformed_first_response_recovers_via_retry(
+        self, engine, trace_store, provider
+    ):
+        """First reply invalid (wrong shape, fenced), second valid → graded.
+
+        This exercises the D-020 layer-4 path INSIDE the engine through the
+        real agents/structured.py — the engine does not reimplement it.
+        """
+        _ingest(trace_store, _complete_trace())
+        provider.replies = [
+            '```json\n{"summary": "wrong shape"}\n```',  # layer 3 rejects
+            json.dumps(RubricJSON.PAYLOAD),               # retry recovers
+        ]
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert provider.calls == 2  # bounded to exactly one retry (layer 4)
+        assert record.verdict == "GRADED"
+        assert record.scores == RubricJSON.PAYLOAD
+        # the retry feedback must carry the validation error back
+        retry_request = provider.requests[1]
+        retry_text = " ".join(m.content for m in retry_request)
+        assert "previous response was invalid" in retry_text
+
+    async def test_persistently_malformed_raises_after_bounded_retry(
+        self, engine, trace_store, provider
+    ):
+        from ai_service.agents.structured import StructuredOutputError
+
+        _ingest(trace_store, _complete_trace())
+        provider.replies = ["not json at all", "still not json"]
+
+        with pytest.raises(StructuredOutputError):
+            await engine.grade("engine-learner", "engine-task")
+        assert provider.calls == 2  # bounded: never more than one retry
+
+
+class TestRegrade:
+    async def test_regrade_overwrites_stored_record(
+        self, engine, trace_store, grade_store, provider
+    ):
+        _ingest(trace_store, _complete_trace())
+
+        first = await engine.grade("engine-learner", "engine-task")
+        assert first.verdict == "GRADED"
+
+        # second grade: the provider now scripts a DIFFERENT rubric outcome
+        updated = json.loads(json.dumps(RubricJSON.PAYLOAD))
+        updated["criteria"]["process_quality"] = 1
+        updated["verdict"] = "not_yet"
+        provider.replies = [json.dumps(updated)]
+
+        second = await engine.grade("engine-learner", "engine-task")
+
+        assert second.scores == updated
+        assert second.created_at >= first.created_at
+        # upsert, not append: exactly one row, holding the LATEST grade
+        grades = grade_store.list_for_learner("engine-learner")
+        assert len(grades) == 1
+        assert grades[0].scores == updated
+        assert grades[0].verdict == "GRADED"
+
+    async def test_regrade_after_gate_outcome_replaces_it(
+        self, engine, trace_store, grade_store, provider, integrity
+    ):
+        """A later successful regrade wholesale-replaces a gate record —
+        the GradeStore upsert contract applied across verdict kinds."""
+        # first: gapped → gate record persisted
+        events = _complete_trace()
+        _ingest(trace_store, [e for e in events if e.seq != 2])
+        gated = await engine.grade("engine-learner", "engine-task")
+        assert gated.verdict == "UNGRADABLE_TRACE_INCOMPLETE"
+
+        # gap healed (late delivery): same pair regrades cleanly
+        trace_store.append(next(e for e in events if e.seq == 2))
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "GRADED"
+        stored = grade_store.get("engine-learner", "engine-task")
+        assert stored is not None
+        assert stored.verdict == "GRADED"
+        assert stored.scores == RubricJSON.PAYLOAD
+        assert len(grade_store.list_for_learner("engine-learner")) == 1
+
+
+class TestBoundary:
+    def test_engine_module_imports_no_fastapi(self):
+        """D-027: the engine module must not import fastapi (transitively
+        beyond the sanctioned agents/structured + llm + telemetry + prompts)."""
+        import sys
+
+        import ai_service.grading.engine as engine_mod
+
+        code = open(engine_mod.__file__).read()
+        assert "fastapi" not in code
+        assert "from ..api" not in code and "from ..api." not in code
+        # the one sanctioned agents import is the module-direct structured defense
+        assert "from ..agents.structured import" in code
+        assert "from ..agents import" not in code
+        # sanity: the module really is loaded (no import cycle surprises)
+        assert engine_mod.__name__ in sys.modules
+
+
+class TestModelValidationSmoke:
+    """RubricScore as a plain pydantic model (D-020 layer-3 target type)."""
+
+    def test_rubric_score_is_frozen_shape_for_scores_dict(self):
+        score = RubricScore.model_validate(RubricJSON.PAYLOAD)
+        dumped = score.model_dump()
+        assert set(dumped["criteria"]) == {
+            "process_quality",
+            "correctness",
+            "debugging_discipline",
+            "test_usage",
+        }
+        assert isinstance(score, BaseModel)
+
+
+class TestVariantAnchorsShipment:
+    """Phase 4 MH#4: variant anchors + seed ship into grading (a-5 same bar)."""
+
+    @pytest.fixture
+    def variant_engine(self, trace_store, grade_store, integrity, provider, tmp_path):
+        store = SQLiteVariantStore(db_path=tmp_path / "variants.db")
+        yield GradingEngine(
+            trace_store,
+            grade_store,
+            integrity,
+            provider,
+            model="gemma4:31b",
+            variant_store=store,
+        ), store
+        store.close()
+
+    @pytest.fixture
+    def plain_engine(self, trace_store, grade_store, integrity, provider):
+        return GradingEngine(
+            trace_store, grade_store, integrity, provider, model="gemma4:31b"
+        )
+
+    async def test_variant_task_grades_with_anchors_and_seed(
+        self, variant_engine, trace_store, provider
+    ):
+        engine, vstore = variant_engine
+        _ingest(trace_store, _complete_trace())
+
+        # A stored variant whose task_id matches the graded trace.
+        from datetime import UTC
+        from datetime import datetime as dt
+
+        from ai_service.variants.store import VariantRecord
+
+        vstore.save(
+            VariantRecord(
+                learner_id="engine-learner",
+                task_id="engine-task",
+                template_id="tpl-llm-judge",
+                seed="cafe" * 16,
+                params={"domain": "tutoring"},
+                statement="Scripted statement long enough to be legal.",
+                starter_files={"README.md": "x"},
+                created_at=dt.now(UTC),
+            )
+        )
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "GRADED"
+        assert record.variant_seed == "cafe" * 16  # D-029 stamped
+        user_msgs = [m.content for req in provider.requests for m in req if m.role == "user"]
+        assert any("Expected effort envelope" in u for u in user_msgs)
+        assert any("tpl-llm-judge" in u for u in user_msgs)
+        assert any("expected_min_test_runs" in u for u in user_msgs)
+
+    async def test_non_variant_task_has_no_anchors(
+        self, variant_engine, trace_store, provider
+    ):
+        engine, _ = variant_engine
+        _ingest(trace_store, _complete_trace())
+
+        record = await engine.grade("engine-learner", "engine-task")
+
+        assert record.verdict == "GRADED"
+        assert record.variant_seed is None
+        user_msgs = [m.content for req in provider.requests for m in req if m.role == "user"]
+        assert not any("Expected effort envelope" in u for u in user_msgs)
+
+    async def test_plain_engine_stays_variant_blind(
+        self, plain_engine, trace_store
+    ):
+        _ingest(trace_store, _complete_trace())
+        record = await plain_engine.grade("engine-learner", "engine-task")
+        assert record.verdict == "GRADED"
+        assert record.variant_seed is None

apps/ai-service/tests/grading/test_features.py

@@ -0,0 +1,170 @@
+"""Trace digest tests (Task 3-1-01, REQ-3-004).
+
+Contract: `compute_digest` is deterministic, bounded (< 4 KB), and carries
+NO raw trace material (commands, file contents, payload strings) — the raw
+trace never reaches the LLM (D-028).
+"""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime, timedelta
+
+import pytest
+
+from ai_service.grading.features import TraceDigest, compute_digest
+from ai_service.telemetry.models import TelemetryEvent
+
+T0 = datetime(2026, 9, 12, 1, 0, 0, tzinfo=UTC)
+
+
+def _event(
+    seq: int, kind: str, payload: dict | None = None, offset_s: float = 0.0
+) -> TelemetryEvent:
+    return TelemetryEvent(
+        learner_id="features-learner",
+        task_id="features-task",
+        seq=seq,
+        kind=kind,
+        payload=payload or {},
+        ts=T0 + timedelta(seconds=offset_s),
+        sandbox_id="sbx-features",
+    )
+
+
+def _paste_and_run_trace() -> list[TelemetryEvent]:
+    """One big file dump, a single passing test at the very end (no cycles)."""
+    return [
+        _event(0, "activity", {"state": "starting"}, 0.0),
+        _event(1, "file_diff", {"path": "main.py", "added": 180, "removed": 0}, 5.0),
+        _event(2, "command", {"cmd": "python -m pytest -q"}, 30.0),
+        _event(3, "test_result", {"passed": True, "exit_code": 0}, 45.0),
+        _event(4, "activity", {"state": "idle"}, 50.0),
+    ]
+
+
+def _iterative_trace() -> list[TelemetryEvent]:
+    """Many small edits; failed runs interleaved; eventual pass (>= 2 cycles)."""
+    events: list[TelemetryEvent] = [
+        _event(0, "activity", {"state": "starting"}, 0.0),
+        _event(1, "file_diff", {"path": "a.py", "added": 12}, 10.0),
+        _event(2, "command", {"cmd": "pytest -q"}, 20.0),
+        _event(3, "test_result", {"passed": False, "exit_code": 1}, 25.0),
+        _event(4, "file_diff", {"path": "a.py", "added": 4, "removed": 2}, 40.0),
+        _event(5, "file_diff", {"path": "b.py", "added": 6}, 50.0),
+        _event(6, "command", {"cmd": "pytest -q"}, 60.0),
+        _event(7, "test_result", {"passed": False, "exit_code": 2}, 65.0),
+        _event(8, "file_diff", {"path": "a.py", "added": 3, "removed": 1}, 80.0),
+        _event(9, "command", {"cmd": "pytest -q tests/"}, 90.0),
+        _event(10, "test_result", {"passed": True, "exit_code": 0}, 95.0),
+        _event(11, "activity", {"state": "idle"}, 100.0),
+    ]
+    return events
+
+
+def test_paste_and_run_vs_iterative_produce_observably_different_digests() -> None:
+    paste = compute_digest(_paste_and_run_trace())
+    iterative = compute_digest(_iterative_trace())
+    assert paste.error_fix_cycles == 0
+    assert iterative.error_fix_cycles == 2
+    assert paste.test_fail_count == 0
+    assert iterative.test_fail_count == 2
+    assert paste.edit_count < iterative.edit_count
+    assert iterative.first_test_pass_offset_s is not None
+    assert paste.first_test_pass_offset_s is not None
+    # Iterative debugs longer before the first pass.
+    assert iterative.first_test_pass_offset_s > paste.first_test_pass_offset_s
+
+
+def test_digest_is_deterministic() -> None:
+    trace = _iterative_trace()
+    assert compute_digest(trace) == compute_digest(list(reversed(trace)))  # seq sort normalizes
+
+
+def test_empty_trace_yields_valid_zeroed_digest() -> None:
+    digest = compute_digest([])
+    assert digest.event_count == 0
+    assert digest.final_test_status == "none"
+    assert digest.first_test_pass_offset_s is None
+    assert digest.mean_fix_latency_s is None
+    assert isinstance(digest, TraceDigest)
+
+
+def test_digest_json_is_bounded_under_4kb() -> None:
+    big = [
+        _event(i, "command", {"cmd": f"grep PATTERN-{i} file-{i}.py"}, i * 1.0)
+        for i in range(200)
+    ]
+    serialized = compute_digest(big).model_dump_json()
+    assert len(serialized.encode()) < 4096, f"digest too large: {len(serialized)}B"
+
+
+def test_no_raw_command_string_leaks_into_digest() -> None:
+    marker = "SECRET-COMMAND-MARKER-7f3a"
+    trace = [
+        _event(0, "command", {"cmd": f"echo {marker} && cat /etc/hostname"}, 0.0),
+        _event(1, "file_diff", {"path": marker + ".py"}, 1.0),
+        _event(2, "run_result", {"exit_code": 0, "stdout": marker}, 2.0),
+    ]
+    digest_json = compute_digest(trace).model_dump_json()
+    assert marker not in digest_json, "raw payload material leaked into digest"
+
+
+def test_idle_gaps_computed_over_threshold() -> None:
+    trace = [
+        _event(0, "activity", {"state": "starting"}, 0.0),
+        _event(1, "activity", {"state": "idle"}, 400.0),  # > 120s gap
+        _event(2, "activity", {"state": "idle"}, 500.0),   # 100s gap (below)
+        _event(3, "activity", {"state": "stopped"}, 800.0),  # > 120s gap
+    ]
+    digest = compute_digest(trace)
+    assert digest.idle_gap_count == 2
+    assert digest.idle_gap_total_s == pytest.approx(400.0 + 300.0, rel=1e-6)
+
+
+def test_command_category_histogram() -> None:
+    trace = [
+        _event(0, "command", {"cmd": "npm run build"}, 0.0),
+        _event(1, "command", {"cmd": "pytest -q"}, 1.0),
+        _event(2, "command", {"cmd": "ls -la"}, 2.0),
+        _event(3, "command", {"cmd": "rm -rf build/"}, 3.0),
+        _event(4, "command", {"cmd": "curl localhost:8420/health"}, 4.0),
+        _event(5, "command", {"cmd": "python mystery.py"}, 5.0),
+    ]
+    digest = compute_digest(trace)
+    assert digest.command_categories == {
+        "build": 1,
+        "debug": 1,
+        "file": 1,
+        "nav": 1,
+        "other": 1,
+        "test": 1,
+    }
+
+
+def test_daemon_topology_mix_is_tolerated() -> None:
+    """P2-verify P1: live traces carry activity+file_diff only — no crash."""
+    trace = [
+        _event(0, "activity", {"state": "starting"}, 0.0),
+        _event(1, "file_diff", {"path": "made-by-exec.txt", "added": 1}, 2.0),
+        _event(2, "activity", {"state": "idle"}, 3.0),
+    ]
+    digest = compute_digest(trace)
+    assert digest.final_test_status == "none"
+    assert digest.edit_count == 1
+    assert digest.test_pass_count == 0
+    assert digest.kind_histogram.get("file_diff") == 1
+
+
+def test_run_result_exit_codes_fall_back_for_test_status() -> None:
+    """No test_result events: run_result exit codes decide pass/fail."""
+    trace = [
+        _event(0, "file_diff", {"path": "x.py"}, 0.0),
+        _event(1, "run_result", {"exit_code": 1}, 10.0),
+        _event(2, "file_diff", {"path": "x.py"}, 20.0),
+        _event(3, "run_result", {"exit_code": 0}, 30.0),
+    ]
+    digest = compute_digest(trace)
+    assert digest.test_fail_count == 1
+    assert digest.test_pass_count == 1
+    assert digest.final_test_status == "pass"
+    assert digest.error_fix_cycles == 1

apps/ai-service/tests/grading/test_store.py

@@ -0,0 +1,268 @@
+"""SQLiteGradeStore tests (REQ-3-004, D-027).
+
+Each test gets its own tmp-path SQLite file — no shared disk state. Covers:
+    - save / get / list_for_learner roundtrip (all fields survive,
+      including nested JSON dicts and the tz-aware created_at contract)
+    - upsert-on-regrade: a second save with the same (learner_id, task_id)
+      REPLACES the row wholesale — scores, verdict, created_at, digest,
+      model and variant_seed all reflect the latest save (documented
+      contract; deliberately opposite of TraceStore.append's dedup)
+    - unknown (learner, task) pair -> None; unknown learner -> empty list
+    - scoping: grades for other learners/tasks are never returned
+    - variant_seed stays None until P4 (D-029) and survives a roundtrip
+    - rows are detached: usable after the store is closed
+    - WAL + synchronous=NORMAL pragmas actually applied to the DB file
+    - concurrent writer + reader against the same DB file (a-3 smoke test)
+"""
+
+import concurrent.futures
+import threading
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+from typing import Any
+
+import pytest
+import sqlalchemy as sa
+
+from ai_service.grading.store import GradeRecord, SQLiteGradeStore
+
+_BASE_TS = datetime(2026, 9, 12, 12, 0, 0, tzinfo=UTC)
+
+
+def make_grade(
+    task_id: str = "task-1",
+    learner_id: str = "learner-1",
+    variant_seed: str | None = None,
+    digest: dict[str, Any] | None = None,
+    scores: dict[str, Any] | None = None,
+    verdict: str = "STRONG",
+    model: str = "gemma4:31b",
+    created_at: datetime | None = None,
+) -> GradeRecord:
+    """Canonical kwargs builder — tests override only what they assert on."""
+    return GradeRecord(
+        learner_id=learner_id,
+        task_id=task_id,
+        variant_seed=variant_seed,
+        digest=digest
+        if digest is not None
+        else {"error_fix_cycles": 3, "command_categories": {"build": 2}},
+        scores=scores
+        if scores is not None
+        else {
+            "criteria": {"process_quality": 3, "correctness": 4},
+            "strengths": ["iterative debugging"],
+            "gaps": ["no final test pass"],
+        },
+        verdict=verdict,
+        model=model,
+        created_at=created_at if created_at is not None else _BASE_TS,
+    )
+
+
+@pytest.fixture
+def store(tmp_path: Path) -> SQLiteGradeStore:
+    s = SQLiteGradeStore(db_path=tmp_path / "grades.db")
+    yield s
+    s.close()
+
+
+def test_save_and_get_roundtrip(store: SQLiteGradeStore) -> None:
+    grade = make_grade()
+    store.save(grade)
+
+    fetched = store.get("learner-1", "task-1")
+    assert fetched is not None
+    assert fetched.learner_id == "learner-1"
+    assert fetched.task_id == "task-1"
+    assert fetched.variant_seed is None  # null until P4 (D-029)
+    assert fetched.digest == grade.digest
+    assert fetched.scores == grade.scores
+    assert fetched.verdict == "STRONG"
+    assert fetched.model == "gemma4:31b"
+    assert fetched.created_at == _BASE_TS
+    assert fetched.created_at.tzinfo is UTC  # tz-normalized on read
+
+
+def test_get_unknown_pair_returns_none(store: SQLiteGradeStore) -> None:
+    store.save(make_grade())
+
+    assert store.get("learner-1", "task-missing") is None
+    assert store.get("learner-missing", "task-1") is None
+    assert store.get("nobody", "nothing") is None
+
+
+def test_list_for_learner_roundtrip(store: SQLiteGradeStore) -> None:
+    # Created out of insertion order; list must come back chronological.
+    store.save(make_grade(task_id="task-c", created_at=_BASE_TS + timedelta(hours=2)))
+    store.save(make_grade(task_id="task-a", created_at=_BASE_TS))
+    store.save(make_grade(task_id="task-b", created_at=_BASE_TS + timedelta(hours=1)))
+
+    grades = store.list_for_learner("learner-1")
+    assert [g.task_id for g in grades] == ["task-a", "task-b", "task-c"]
+    assert all(g.learner_id == "learner-1" for g in grades)
+    hours = (timedelta(hours=0), timedelta(hours=1), timedelta(hours=2))
+    assert all(
+        g.created_at == _BASE_TS + offset
+        for g, offset in zip(grades, hours, strict=True)
+    )
+    assert all(g.created_at.tzinfo is UTC for g in grades)
+
+
+def test_list_for_learner_unknown_learner_returns_empty_list(
+    store: SQLiteGradeStore,
+) -> None:
+    assert store.list_for_learner("nobody") == []
+
+
+def test_lists_are_scoped_to_the_learner(store: SQLiteGradeStore) -> None:
+    store.save(make_grade(learner_id="learner-1", task_id="task-1"))
+    store.save(make_grade(learner_id="learner-2", task_id="task-1"))
+
+    assert [g.task_id for g in store.list_for_learner("learner-1")] == ["task-1"]
+    assert [g.learner_id for g in store.list_for_learner("learner-2")] == ["learner-2"]
+    # (learner-2, task-1) is a distinct row: same task_id, different grade.
+    grades_2 = store.list_for_learner("learner-2")
+    assert len(grades_2) == 1
+    assert grades_2[0].learner_id == "learner-2"
+
+
+def test_regrade_overwrites_the_stored_row(store: SQLiteGradeStore) -> None:
+    """THE contract of this store (upsert on the PK pair, latest wins).
+
+    The grading engine re-grades a task as its rubric or input evolves;
+    the second save replaces scores, verdict, created_at, digest, model
+    and variant_seed wholesale — exactly one row survives per pair.
+    """
+    first = make_grade(
+        verdict="DEVELOPING",
+        model="gemma4:31b",
+        scores={"criteria": {"process_quality": 1}},
+        digest={"error_fix_cycles": 0},
+        created_at=_BASE_TS,
+    )
+    store.save(first)
+
+    second = make_grade(
+        verdict="EXEMPLARY",
+        model="gemma4:31b-p2",
+        variant_seed="seed-77",
+        scores={"criteria": {"process_quality": 4}},
+        digest={"error_fix_cycles": 6},
+        created_at=_BASE_TS + timedelta(hours=1),
+    )
+    store.save(second)
+
+    fetched = store.get("learner-1", "task-1")
+    assert fetched is not None
+    # The regrade replaced every field of the first save.
+    assert fetched.verdict == "EXEMPLARY"
+    assert fetched.model == "gemma4:31b-p2"
+    assert fetched.variant_seed == "seed-77"
+    assert fetched.scores == {"criteria": {"process_quality": 4}}
+    assert fetched.digest == {"error_fix_cycles": 6}
+    assert fetched.created_at == _BASE_TS + timedelta(hours=1)
+
+    # Latest-wins also holds in list_for_learner — one row, not two.
+    grades = store.list_for_learner("learner-1")
+    assert len(grades) == 1
+    assert grades[0].verdict == "EXEMPLARY"
+
+
+def test_regrade_preserves_other_pairs(store: SQLiteGradeStore) -> None:
+    # An upsert on (learner-1, task-1) must not touch (learner-1, task-2).
+    store.save(make_grade(task_id="task-1", verdict="STRONG"))
+    store.save(make_grade(task_id="task-2", verdict="DEVELOPING"))
+    store.save(make_grade(task_id="task-1", verdict="EXEMPLARY"))
+
+    other = store.get("learner-1", "task-2")
+    assert other is not None
+    assert other.verdict == "DEVELOPING"  # untouched by the task-1 regrade
+    assert len(store.list_for_learner("learner-1")) == 2
+
+
+def test_empty_scores_dict_roundtrips(store: SQLiteGradeStore) -> None:
+    # Legal shape: an UNGRADABLE_TRACE_INCOMPLETE record carries a verdict
+    # but no scores (and here, no digest either).
+    store.save(make_grade(verdict="UNGRADABLE_TRACE_INCOMPLETE", scores={}, digest={}))
+
+    fetched = store.get("learner-1", "task-1")
+    assert fetched is not None
+    assert fetched.scores == {}
+    assert fetched.digest == {}
+    assert fetched.verdict == "UNGRADABLE_TRACE_INCOMPLETE"
+
+
+def test_rows_are_detached_after_save(store: SQLiteGradeStore, tmp_path: Path) -> None:
+    # The engine hands GradeRecords across layers; rows must survive the
+    # store that produced them being closed (no open-session ORM magic).
+    store.save(make_grade())
+    fetched = store.get("learner-1", "task-1")
+    store.close()
+
+    assert fetched is not None
+    assert fetched.verdict == "STRONG"
+    assert fetched.scores["criteria"]["process_quality"] == 3
+
+    # A fresh store on the same file sees the same row (durability).
+    reopened = SQLiteGradeStore(db_path=tmp_path / "grades.db")
+    try:
+        again = reopened.get("learner-1", "task-1")
+        assert again is not None
+        assert again.verdict == "STRONG"
+    finally:
+        reopened.close()
+
+
+def test_pragmas_are_applied(store: SQLiteGradeStore) -> None:
+    # Pragmas are per-connection; query through the store's engine so the
+    # connect hook (not a default sqlite3 connection) is what we inspect.
+    with store._engine.connect() as conn:
+        (journal_mode,) = conn.execute(sa.text("PRAGMA journal_mode")).one()
+        (synchronous,) = conn.execute(sa.text("PRAGMA synchronous")).one()
+    assert journal_mode == "wal"
+    # synchronous=NORMAL is 1 in SQLite's pragma numbering.
+    assert synchronous == 1
+
+
+def test_concurrent_writer_and_reader_no_database_is_locked(tmp_path: Path) -> None:
+    """One thread saves while another reads in a tight loop (a-3).
+
+    Without WAL + busy_timeout this pattern reliably produces
+    `OperationalError: database is locked` on SQLite. The assertion is
+    that every reader call completes and the latest write lands intact.
+    """
+    db_path = tmp_path / "grades.db"
+    n_pairs = 60  # distinct tasks -> distinct PK pairs, one regrade each
+    stop_writing = threading.Event()
+
+    writer = SQLiteGradeStore(db_path=db_path)
+    reader = SQLiteGradeStore(db_path=db_path)
+    try:
+
+        def write_grades() -> None:
+            for seq in range(n_pairs):
+                writer.save(
+                    make_grade(
+                        task_id=f"task-{seq}",
+                        created_at=_BASE_TS + timedelta(seconds=seq),
+                    )
+                )
+            stop_writing.set()
+
+        def read_grades() -> None:
+            while not stop_writing.is_set():
+                reader.list_for_learner("learner-1")
+            # Final read after the writer is done.
+            assert len(reader.list_for_learner("learner-1")) == n_pairs
+
+        with concurrent.futures.ThreadPoolExecutor(max_workers=2) as pool:
+            futures = [pool.submit(write_grades), pool.submit(read_grades)]
+            for future in futures:
+                future.result(timeout=30)
+
+        # Every save was an upsert on its own pair: nothing lost, none doubled.
+        assert len(writer.list_for_learner("learner-1")) == n_pairs
+    finally:
+        reader.close()
+        writer.close()

apps/ai-service/tests/sandbox/test_sandbox_agent.py

@@ -0,0 +1,632 @@
+"""Unit tests for scripts/sandbox-agent.py (Task 2-2-01, REQ-3-003).
+
+Live-agent tests run the real agent (no sandbox namespace required — it is a
+plain stdlib process) against a loopback fake WebSocket server implemented
+with raw socket frames, mirroring the agent's client codec. Covers:
+
+  * ordered seq emission over a live connection;
+  * spool-on-disconnect (socket drop leaves events durable in the spool);
+  * reconnect + flush preserves order (no loss; one-line replay margin may
+    duplicate the last pre-drop event — the server dedups on
+    (learner, task, seq));
+  * SIGKILL resilience (boot replays the spool, `seq` resumes monotonically);
+  * file_diff polling capture (created / modified / deleted);
+  * an AST scan asserting the agent file is stdlib-only (D-031).
+"""
+
+from __future__ import annotations
+
+import ast
+import base64
+import hashlib
+import json
+import socket
+import struct
+import sys
+import threading
+import time
+from collections.abc import Iterator
+from datetime import datetime
+from importlib.util import module_from_spec, spec_from_file_location
+from pathlib import Path
+
+import pytest
+
+AGENT_PATH = (
+    Path(__file__).resolve().parents[2] / "scripts" / "sandbox-agent.py"
+).resolve()
+
+_WS_GUID = "258EAFA5-E914-47DA-95CA-C5AB0DC85B11"
+
+# Python 3.11.2+ stdlib top-level names (incl. platform leaves removed in 3.13
+# — the AST scan asserts policy, i.e. no third-party imports, not 3.13 import
+# viability). Kept explicit because sys.stdlib_module_names lacks them.
+_PY311_EXTRA_STDLIB = {
+    "aifc",
+    "asynchat",
+    "asyncore",
+    "audioop",
+    "cgi",
+    "cgitb",
+    "chunk",
+    "crypt",
+    "imghdr",
+    "mailcap",
+    "msilib",
+    "nis",
+    "nntplib",
+    "ossaudiodev",
+    "pipes",
+    "sndhdr",
+    "spwd",
+    "sunau",
+    "uu",
+    "xdrlib",
+    "__future__",
+}
+_ALLOWED_MODULES = frozenset(set(sys.stdlib_module_names) | _PY311_EXTRA_STDLIB)
+
+
+def _load_agent():
+    """Load the hyphenated script as a module (it is not importable by name)."""
+    spec = spec_from_file_location("sandbox_agent", AGENT_PATH)
+    assert spec is not None and spec.loader is not None
+    mod = module_from_spec(spec)
+    sys.modules[spec.name] = mod
+    spec.loader.exec_module(mod)
+    return mod
+
+
+agent = _load_agent()
+
+# ---------------------------------------------------------------------------
+# Raw-socket frame helpers (test-harness mirror of the agent's codec)
+# ---------------------------------------------------------------------------
+
+
+def _read_exact(conn: socket.socket, n: int) -> bytes:
+    buf = bytearray()
+    while len(buf) < n:
+        chunk = conn.recv(n - len(buf))
+        if not chunk:
+            raise ConnectionError("unexpected EOF")
+        buf += chunk
+    return bytes(buf)
+
+
+def _server_read_frame(conn: socket.socket) -> tuple[int, bytes]:
+    """Read one client frame (client frames are always masked per RFC 6455)."""
+    b0, b1 = _read_exact(conn, 2)
+    opcode = b0 & 0x0F
+    length = b1 & 0x7F
+    if length == 126:
+        length = struct.unpack("!H", _read_exact(conn, 2))[0]
+    elif length == 127:
+        length = struct.unpack("!Q", _read_exact(conn, 8))[0]
+    mask = _read_exact(conn, 4) if (b1 & 0x80) else b""
+    payload = _read_exact(conn, length) if length else b""
+    if mask:
+        payload = bytes(b ^ mask[i % 4] for i, b in enumerate(payload))
+    return opcode, payload
+
+
+def _server_send_frame(conn: socket.socket, opcode: int, payload: bytes = b"") -> None:
+    header = bytearray([0x80 | opcode])
+    n = len(payload)
+    if n < 126:
+        header.append(n)
+    elif n < 65536:
+        header.append(126)
+        header += struct.pack("!H", n)
+    else:
+        header.append(127)
+        header += struct.pack("!Q", n)
+    conn.sendall(bytes(header) + payload)
+
+
+# ---------------------------------------------------------------------------
+# Loopback fake WS ingest server (thread-per-connection, no dependencies)
+# ---------------------------------------------------------------------------
+
+
+class _WSConn:
+    def __init__(self, sock: socket.socket) -> None:
+        self.sock = sock
+        self.alive = True
+
+
+class FakeWSServer:
+    """Accepts RFC 6455 handshakes and collects text frames in arrival order."""
+
+    def __init__(self) -> None:
+        self._listener = socket.socket()
+        self._listener.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+        self._listener.bind(("127.0.0.1", 0))
+        self._listener.listen(8)
+        self._listener.settimeout(0.5)
+        self.port: int = self._listener.getsockname()[1]
+        self._connections: list[_WSConn] = []
+        self._frames: list[tuple[int, bytes]] = []
+        self._lock = threading.Lock()
+        self._stop = threading.Event()
+        self._thread = threading.Thread(target=self._run, daemon=True)
+
+    @property
+    def url(self) -> str:
+        return f"ws://127.0.0.1:{self.port}/v1/telemetry/ingest"
+
+    @property
+    def events(self) -> list[dict]:
+        with self._lock:
+            return [json.loads(payload) for op, payload in self._frames if op == 0x1]
+
+    def has_live_connection(self) -> bool:
+        with self._lock:
+            return any(c.alive for c in self._connections)
+
+    def count_connections(self) -> int:
+        with self._lock:
+            return len(self._connections)
+
+    def kill_all_connections(self) -> None:
+        """Abruptly close the socket without a WS close frame (network drop)."""
+        with self._lock:
+            conns = list(self._connections)
+        for conn in conns:
+            conn.alive = False
+            try:
+                conn.sock.shutdown(socket.SHUT_RDWR)
+            except OSError:
+                pass
+            try:
+                conn.sock.close()
+            except OSError:
+                pass
+
+    # -- internals ------------------------------------------------------
+    def start(self) -> None:
+        self._thread.start()
+
+    def stop(self) -> None:
+        self._stop.set()
+        self.kill_all_connections()
+        self._thread.join(timeout=3)
+        self._listener.close()
+
+    def _run(self) -> None:
+        while not self._stop.is_set():
+            try:
+                sock, _ = self._listener.accept()
+            except TimeoutError:
+                continue
+            except OSError:
+                return
+            threading.Thread(target=self._serve, args=(sock,), daemon=True).start()
+
+    def _serve(self, sock: socket.socket) -> None:
+        conn = _WSConn(sock)
+        with self._lock:
+            self._connections.append(conn)
+        try:
+            self._handshake(sock)
+            while conn.alive and not self._stop.is_set():
+                opcode, payload = _server_read_frame(sock)
+                if opcode == 0x8:  # close
+                    _server_send_frame(sock, 0x8, payload)
+                    break
+                with self._lock:
+                    self._frames.append((opcode, payload))
+        except (ConnectionError, OSError):
+            pass
+        finally:
+            conn.alive = False
+            try:
+                sock.close()
+            except OSError:
+                pass
+
+    @staticmethod
+    def _handshake(sock: socket.socket) -> None:
+        buf = b""
+        while b"\r\n\r\n" not in buf:
+            chunk = sock.recv(4096)
+            if not chunk:
+                raise ConnectionError("EOF during handshake")
+            buf += chunk
+        headers: dict[str, str] = {}
+        for line in buf.decode("latin-1").split("\r\n")[1:]:
+            if ":" in line:
+                name, _, value = line.partition(":")
+                headers[name.strip().lower()] = value.strip()
+        accept = base64.b64encode(
+            hashlib.sha1((headers["sec-websocket-key"] + _WS_GUID).encode()).digest()
+        ).decode()
+        sock.sendall(
+            (
+                "HTTP/1.1 101 Switching Protocols\r\n"
+                "Upgrade: websocket\r\n"
+                "Connection: Upgrade\r\n"
+                f"Sec-WebSocket-Accept: {accept}\r\n\r\n"
+            ).encode("ascii")
+        )
+
+
+# ---------------------------------------------------------------------------
+# Killable TCP proxy: lets a test hold the network down deterministically.
+# ---------------------------------------------------------------------------
+
+
+class KillableProxy:
+    """Forwards bytes both ways; `kill()` severs + closes the listener so the
+    client can neither talk nor reconnect until `resume()` re-opens it."""
+
+    def __init__(self, target_port: int) -> None:
+        self._target = ("127.0.0.1", target_port)
+        self._listener = socket.socket()
+        self._listener.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+        self._listener.bind(("127.0.0.1", 0))
+        self._listener.listen(8)
+        self._listener.settimeout(0.2)
+        self.port: int = self._listener.getsockname()[1]
+        self._up = threading.Event()
+        self._up.set()
+        self._dead = threading.Event()
+        self._conns: list[socket.socket] = []
+        self._lock = threading.Lock()
+        self._thread = threading.Thread(target=self._run, daemon=True)
+
+    @property
+    def url(self) -> str:
+        return f"ws://127.0.0.1:{self.port}/v1/telemetry/ingest"
+
+    def start(self) -> None:
+        self._thread.start()
+
+    def kill(self) -> None:
+        """Sever live sockets AND stop accepting until resumed (network down)."""
+        self._up.clear()
+        with self._lock:
+            conns = list(self._conns)
+        for s in conns:
+            try:
+                s.shutdown(socket.SHUT_RDWR)
+            except OSError:
+                pass
+            try:
+                s.close()
+            except OSError:
+                pass
+
+    def resume(self) -> None:
+        self._up.set()
+
+    def stop(self) -> None:
+        self._dead.set()
+        self.kill()
+        self._thread.join(timeout=3)
+        self._listener.close()
+
+    def _run(self) -> None:
+        while not self._dead.is_set():
+            try:
+                client, _ = self._listener.accept()
+            except TimeoutError:
+                continue
+            except OSError:
+                return
+            if not self._up.is_set():
+                client.close()
+                continue
+            try:
+                upstream = socket.create_connection(self._target, timeout=2)
+            except OSError:
+                client.close()
+                continue
+            with self._lock:
+                self._conns.extend([client, upstream])
+            threading.Thread(
+                target=self._pump, args=(client, upstream), daemon=True
+            ).start()
+            threading.Thread(
+                target=self._pump, args=(upstream, client), daemon=True
+            ).start()
+
+    def _pump(self, src: socket.socket, dst: socket.socket) -> None:
+        try:
+            while self._up.is_set() and not self._dead.is_set():
+                data = src.recv(65536)
+                if not data:
+                    break
+                dst.sendall(data)
+        except OSError:
+            pass
+        finally:
+            for s in (src, dst):
+                try:
+                    s.shutdown(socket.SHUT_RDWR)
+                except OSError:
+                    pass
+
+
+# ---------------------------------------------------------------------------
+# Fixtures / helpers
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+def fake_server() -> Iterator[FakeWSServer]:
+    server = FakeWSServer()
+    server.start()
+    yield server
+    server.stop()
+
+
+@pytest.fixture
+def held_link(fake_server: FakeWSServer) -> Iterator[KillableProxy]:
+    """A killable proxy in front of the fake server (network-down control)."""
+    proxy = KillableProxy(fake_server.port)
+    proxy.start()
+    yield proxy
+    proxy.stop()
+
+def _config(tmp_path: Path, url: str, spool_path: Path | None = None) -> agent.AgentConfig:
+    workspace = tmp_path / "workspace"
+    return agent.AgentConfig(
+        learner_id="learner-1",
+        task_id="task-1",
+        ingest_url=url,
+        sandbox_id="sb-test",
+        workspace=workspace,
+        spool_path=spool_path or (tmp_path / "spool" / "spool.jsonl"),
+        poll_interval_s=0.05,
+        activity_interval_s=30.0,
+        backoff_base_s=0.05,
+        backoff_max_s=0.2,
+    )
+
+
+def _make_agent(tmp_path: Path, url: str, spool_path: Path | None = None) -> agent.Agent:
+    return agent.Agent(_config(tmp_path, url, spool_path))
+
+
+def _wait_until(pred, timeout_s: float = 8.0, interval_s: float = 0.02) -> bool:
+    deadline = time.monotonic() + timeout_s
+    while time.monotonic() < deadline:
+        if pred():
+            return True
+        time.sleep(interval_s)
+    return pred()
+
+
+# ---------------------------------------------------------------------------
+# Tests
+# ---------------------------------------------------------------------------
+
+
+class TestOrderedEmission:
+    def test_command_produces_ordered_seq_chain(
+        self, tmp_path: Path, fake_server: FakeWSServer
+    ) -> None:
+        test_agent = _make_agent(tmp_path, fake_server.url)
+        test_agent.start()
+        try:
+            assert test_agent.wait_connected(5), "agent never connected"
+            test_agent.run_command("echo hello-agent")
+        finally:
+            test_agent.stop()
+        assert _wait_until(lambda: len(fake_server.events) >= 6)
+
+        events = fake_server.events
+        delivered_seqs = [e["seq"] for e in events]
+        assert delivered_seqs == sorted(delivered_seqs), "wire order must be seq order"
+        assert min(delivered_seqs) == 0
+
+        kinds = [e["kind"] for e in events if e["kind"] != "activity"]
+        assert kinds == ["stdin", "command", "stdout", "run_result"], kinds
+        stdout = next(e for e in events if e["kind"] == "stdout")
+        assert "hello-agent" in stdout["payload"]["data"]
+        run_result = next(e for e in events if e["kind"] == "run_result")
+        assert run_result["payload"]["exit_code"] == 0
+
+        # Wire contract (D-026): identity is bound at the WS handshake (URL
+        # query params) — the frame body must NOT carry learner_id/task_id
+        # (the ingest endpoint rejects them with extra="forbid").
+        required = {"seq", "kind", "payload", "ts"}
+        forbidden = {"learner_id", "task_id"}
+        for event in events:
+            assert required <= set(event), event
+            assert not (forbidden & set(event)), f"URL-owned ids leaked in frame: {event}"
+            assert event.get("sandbox_id", "sb-test") == "sb-test"
+            datetime.fromisoformat(event["ts"])  # must parse
+
+    def test_test_command_classified_as_test_result(
+        self, tmp_path: Path, fake_server: FakeWSServer
+    ) -> None:
+        test_agent = _make_agent(tmp_path, fake_server.url)
+        test_agent.start()
+        try:
+            assert test_agent.wait_connected(5)
+            test_agent.run_command("pytest -q")
+        finally:
+            test_agent.stop()
+        kinds = [e["kind"] for e in fake_server.events]
+        assert "test_result" in kinds
+        assert "run_result" not in kinds
+
+
+class TestSpoolOnDisconnect:
+    def test_disconnect_spools_events_without_loss(
+        self, tmp_path: Path, fake_server: FakeWSServer, held_link: KillableProxy
+    ) -> None:
+        test_agent = _make_agent(tmp_path, held_link.url)
+        test_agent.start()
+        assert test_agent.wait_connected(5)
+        first = test_agent.run_command("echo before-drop")
+        assert _wait_until(
+            lambda: any(
+                e["kind"] == "run_result" and e["seq"] == first["seq"]
+                for e in fake_server.events
+            )
+        )
+
+        held_link.kill()  # network hard-down: sockets severed, listener closed
+        assert test_agent.wait_disconnected(5), "agent never noticed the drop"
+        seq_before = test_agent._seq  # noqa: SLF001 — white-box unit test
+        spooled = [
+            test_agent.emit("activity", {"state": "offline", "n": n}) for n in range(3)
+        ]
+        spooled_seqs = {e["seq"] for e in spooled}
+        assert min(spooled_seqs) == seq_before, "seqs must continue monotonically"
+        test_agent.stop()
+
+        # Network stayed down for the whole window → nothing was delivered.
+        spool_lines = test_agent._spool.read_all()
+        spool_seqs = [json.loads(line)["seq"] for line in spool_lines]
+        assert spooled_seqs <= set(spool_seqs), "dropped events must be spool-durable"
+        received_seqs = [e["seq"] for e in fake_server.events]
+        assert spooled_seqs.isdisjoint(received_seqs), (
+            "events emitted during the outage must NOT have reached the server"
+        )
+        # seq is monotonic across the whole run: offline events + the final
+        # stop marker account for every value allocated after the drop.
+        assert test_agent._seq == seq_before + 4  # 3 offline + "stopped"
+
+
+class TestReconnectFlush:
+    def test_reconnect_flushes_spool_in_order_no_loss(
+        self, tmp_path: Path, fake_server: FakeWSServer, held_link: KillableProxy
+    ) -> None:
+        test_agent = _make_agent(tmp_path, held_link.url)
+        test_agent.start()
+        assert test_agent.wait_connected(5)
+        first = test_agent.run_command("echo first")
+        # P7 de-flake: wait for the pre-kill burst to be OBSERVED at the
+        # server before severing (the sibling TestSpoolOnDisconnect test
+        # already had this discipline). Killing mid-burst exercises a
+        # DIFFERENT, documented limitation — the agent's one-line replay
+        # margin cannot cover a multi-frame TCP in-flight window (an
+        # ACK-protocol gap tracked for v0.4) — which made this test
+        # nondeterministic under load instead of testing what its name
+        # says: the reconnect flush of OFFLINE-spooled events.
+        assert _wait_until(
+            lambda: any(
+                e["kind"] == "run_result" and e["seq"] == first["seq"]
+                for e in fake_server.events
+            ),
+            timeout_s=10.0,
+        ), "pre-kill burst never reached the server"
+
+        held_link.kill()  # outage begins: no traffic, no reconnect possible
+        assert test_agent.wait_disconnected(5)
+        buffered = [test_agent.emit("activity", {"state": "offline", "n": n}) for n in range(3)]
+        buffered_seqs = [e["seq"] for e in buffered]
+        assert set(buffered_seqs).isdisjoint(e["seq"] for e in fake_server.events)
+
+        held_link.resume()  # outage ends → supervisor reconnects, flushes spool
+        # 20s deadline: under full-suite load the supervisor thread can be
+        # starved past its normal sub-second reconnect; 8s was flaky.
+        assert _wait_until(
+            lambda: all(
+                seq in {e["seq"] for e in fake_server.events} for seq in buffered_seqs
+            ),
+            timeout_s=20.0,
+        ), "spooled events never flushed after reconnect"
+        test_agent.stop()
+
+        replayed_wire_seqs = [e["seq"] for e in fake_server.events]
+        # No loss: every emitted seq arrived, in ascending wire order (dedup
+        # unique seqs — the one-line replay margin may repeat the pre-drop
+        # tail event; the server dedups on (learner, task, seq)).
+        unique_seqs = sorted(set(replayed_wire_seqs))
+        assert unique_seqs == list(range(test_agent._seq)), (
+            f"loss or gap: wire had {unique_seqs}, expected 0..{test_agent._seq - 1}"
+        )
+        positions = {seq: replayed_wire_seqs.index(seq) for seq in unique_seqs}
+        assert [positions[s] for s in unique_seqs] == sorted(positions.values())
+        # Dupes (if any) are confined to the replay margin: at or before the
+        # first buffered (outage) seq.
+        duped = {s for s in replayed_wire_seqs if replayed_wire_seqs.count(s) > 1}
+        assert all(s <= min(buffered_seqs) for s in duped), f"unexpected dupes: {duped}"
+
+
+class TestSpoolResumeAfterRestart:
+    def test_restart_recovers_seq_and_pending(self, tmp_path: Path) -> None:
+        """Simulates SIGKILL: no stop(), spool must carry state into next boot."""
+        spool_path = tmp_path / "spool.jsonl"
+        cfg = _config(tmp_path, "ws://127.0.0.1:1/x", spool_path)  # unreachable
+        first = agent.Agent(cfg)
+        e0 = first.emit("activity", {"state": "boot"})
+        e1 = first.emit("stdin", {"line": "ls"})
+        del first  # "killed" — no stop(), nothing flushed beyond per-append fsync
+
+        second = agent.Agent(cfg)
+        try:
+            assert second._seq == 2, f"seq must resume at 2, got {second._seq}"
+            pending_seqs = [json.loads(line)["seq"] for line in second._pending]
+            assert pending_seqs == [e0["seq"], e1["seq"]]
+        finally:
+            second.stop()
+        # stop() with no connection preserves pending events on disk, plus the
+        # final "stopped" activity marker.
+        on_disk = agent.Spool(spool_path).read_all()
+        assert len(on_disk) == 3
+        assert json.loads(on_disk[-1])["payload"]["state"] == "stopped"
+
+
+class TestFileDiffWatcher:
+    def test_workspace_changes_emitted(self, tmp_path: Path) -> None:
+        test_agent = _make_agent(tmp_path, "ws://127.0.0.1:1/x")  # offline: pure capture
+        try:
+            test_agent.start()
+            target = test_agent.config.workspace / "hello.txt"
+            target.write_text("v1\n")
+            assert _wait_until(lambda: _find_diff(test_agent, "hello.txt", "created"))
+            target.write_text("v1\nv2\n")
+            assert _wait_until(lambda: _find_diff(test_agent, "hello.txt", "modified"))
+            target.unlink()
+            assert _wait_until(lambda: _find_diff(test_agent, "hello.txt", "deleted"))
+        finally:
+            test_agent.stop()
+
+        created = _find_diff(test_agent, "hello.txt", "created")
+        assert created["payload"]["diff"]  # unified diff present
+        assert "v1" in created["payload"]["diff"]
+
+        # the spool dir itself must never be reported as a diff
+        paths = [
+            json.loads(line)["payload"].get("path")
+            for line in test_agent._spool.read_all()  # noqa: SLF001
+            if json.loads(line)["kind"] == "file_diff"
+        ]
+        assert all(p and not p.startswith(".nc-") for p in paths)
+
+
+def _find_diff(test_agent: agent.Agent, path: str, change: str) -> dict | None:
+    for line in test_agent._spool.read_all():  # noqa: SLF001
+        event = json.loads(line)
+        if (
+            event["kind"] == "file_diff"
+            and event["payload"].get("path") == path
+            and event["payload"].get("change") == change
+        ):
+            return event
+    return None
+
+
+class TestStdlibOnly:
+    def test_no_third_party_imports(self) -> None:
+        """D-031: the agent runs in-namespace where only stdlib exists."""
+        tree = ast.parse(AGENT_PATH.read_text(encoding="utf-8"))
+        imported: set[str] = set()
+        for node in ast.walk(tree):
+            if isinstance(node, ast.Import):
+                imported.update(alias.name.split(".")[0] for alias in node.names)
+            elif isinstance(node, ast.ImportFrom) and node.level == 0:
+                if node.module:
+                    imported.add(node.module.split(".")[0])
+        third_party = imported - _ALLOWED_MODULES
+        assert not third_party, (
+            f"sandbox-agent.py must be stdlib-only (D-031); found: {sorted(third_party)}"
+        )
+        # sanity: the scan really saw the agent's core imports
+        assert {"socket", "json", "threading", "ssl", "subprocess"} <= imported

apps/ai-service/tests/sandbox/test_telemetry_wiring.py

@@ -0,0 +1,203 @@
+"""End-to-end telemetry wiring (Task 2-3-01, REQ-3-003).
+
+Real-namespace probe: a telemetry-wired sandbox (``create(..., task_id=...)``)
+spawns the stdlib capture agent, which dials a REAL uvicorn server over
+loopback (a TestClient app has no listening socket, so a true subprocess
+cannot reach it — this test must boot the actual ASGI server). Commands
+executed via the sandbox exec path produce ordered telemetry events that
+arrive at the live WS ingest endpoint and persist in SQLite.
+
+Probe-guarded: skips (not fails) on hosts without user namespaces.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import contextlib
+import socket
+import time
+from pathlib import Path
+
+import pytest
+import uvicorn
+
+from ai_service.config import Settings
+from ai_service.main import create_app
+from ai_service.sandbox import SandboxManager
+from ai_service.sandbox.unshare_backend import UnshareBackend
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+
+from .test_isolation import USERSNS_AVAILABLE
+
+
+def _userns_probe() -> None:
+    if not USERSNS_AVAILABLE:
+        pytest.skip("user namespaces unavailable on this host (probe)")
+
+
+def _free_port() -> int:
+    with socket.socket() as s:
+        s.bind(("127.0.0.1", 0))
+        return s.getsockname()[1]
+
+
+@pytest.mark.asyncio
+async def test_task_sandbox_streams_events_to_ingest(tmp_path: Path) -> None:
+    """create(task_id=...) -> exec -> events land in SQLite in order (e2e)."""
+    _userns_probe()
+
+    learner = "wiring-learner"
+    task = "task-e2e-1"
+    db = tmp_path / "wiring.db"
+    store = SQLiteTraceStore(db_path=db)
+    app = create_app()
+    app.state.trace_store = store
+    app.state.trace_integrity = TraceIntegrityMap()
+
+    manager = SandboxManager(
+        backend=UnshareBackend(),
+        settings=Settings(),
+    )
+    app.state.sandbox_manager = manager
+
+    port = _free_port()
+    config = uvicorn.Config(app, host="127.0.0.1", port=port, log_level="warning")
+    server = uvicorn.Server(config)
+    serve_task = asyncio.get_running_loop().create_task(server.serve())
+    try:
+        deadline = time.monotonic() + 10.0
+        while not server.started and time.monotonic() < deadline:
+            await asyncio.sleep(0.05)
+        assert server.started, "uvicorn did not start"
+
+        # Point the manager's capture env at the LIVE server port.
+        manager._settings = Settings(telemetry_ingest_host="127.0.0.1")  # noqa: SLF001
+        orig_capture_env = manager._capture_env  # noqa: SLF001
+
+        def _capture_env(sandbox_id: str, learner_id: str, task_id: str):
+            env = orig_capture_env(sandbox_id, learner_id, task_id)
+            env["NC_INGEST_URL"] = (
+                f"ws://127.0.0.1:{port}/v1/telemetry/ingest"
+                f"?learner_id={learner}&task_id={task}&sandbox_id={sandbox_id}"
+            )
+            return env
+
+        manager._capture_env = _capture_env  # noqa: SLF001
+
+        handle = await manager.create(learner, task_id=task)
+        try:
+            live = manager._handles[handle.id]  # noqa: SLF001
+            result = await manager._backend.exec(  # noqa: SLF001
+                live, ["sh", "-c", "echo hello-telemetry && echo second-line"]
+            )
+            assert result.returncode == 0, result.stderr
+
+            # In the persistent-sandbox topology the agent observes execs
+            # through its workspace watcher (the exec path is a separate
+            # nsenter join), so the streaming kinds are activity + file_diff;
+            # command/stdout kinds belong to the agent REPL (interactive use).
+            events = await _await_events(store, learner, task, minimum=2)
+            seqs = [e.seq for e in events]
+            assert seqs == sorted(seqs), f"events out of order: {seqs}"
+            kinds = {e.kind for e in events}
+            assert kinds & {"activity", "file_diff", "command"}, kinds
+            assert all(e.sandbox_id == handle.id for e in events), "sandbox_id mismatch"
+
+            # The learner path stays scoped: nothing for a different task.
+            assert store.get_trace(learner, "some-other-task") == []
+        finally:
+            await manager.destroy(handle.id)
+    finally:
+        server.should_exit = True
+        with contextlib.suppress(Exception):
+            await asyncio.wait_for(serve_task, timeout=10.0)
+        store.close()
+
+
+@pytest.mark.asyncio
+async def test_no_task_id_means_no_capture(tmp_path: Path) -> None:
+    """Pure shell sandbox (task_id=None) spawns no capture agent (REQ-3-001 path)."""
+    _userns_probe()
+
+    store = SQLiteTraceStore(db_path=tmp_path / "shell.db")
+    backend = UnshareBackend()
+    manager = SandboxManager(backend=backend, settings=Settings())
+
+    handle = await manager.create("shell-learner")
+    try:
+        assert handle.id not in backend._tracked  # noqa: SLF001
+        live = manager._handles[handle.id]  # noqa: SLF001
+        result = await backend.exec(live, ["sh", "-c", "echo plain"])
+        assert "plain" in result.stdout
+    finally:
+        await manager.destroy(handle.id)
+    store.close()
+
+
+@pytest.mark.asyncio
+async def test_destroy_kills_inner_namespace_not_just_the_shim() -> None:
+    """Destroy must reap the ns-init, not only the `unshare --fork` shim.
+
+    Regression: `_reap(inner)` kills the unshare PARENT, but its forked child
+    (PID 1 of the sandbox pid/mnt/net namespace, the `sleep`) reparents to
+    host init and holds the tmpfs + workspace bind for the full sleep
+    duration — a leaked sandbox per destroy (observed ~30 orphaned
+    `sleep 3600` processes after one suite run; `--kill-child` did not reach
+    it under this flag combo). The ns-init's host pid is `tracked.inner_pid`.
+    """
+    _userns_probe()
+
+    backend = UnshareBackend()
+    manager = SandboxManager(backend=backend, settings=Settings())
+
+    handle = await manager.create("lifecycle-learner", task_id="lifecycle-task")
+    try:
+        tracked = backend._tracked[handle.id]  # noqa: SLF001
+        assert tracked.agent is not None
+        assert _proc_alive(tracked.agent.pid), "agent should live with the sandbox"
+        assert _proc_alive(tracked.inner_pid), "ns-init should live with the sandbox"
+    finally:
+        await manager.destroy(handle.id)
+
+    # Reaping the unshare shim is asynchronous up to _reap's timeouts; poll
+    # until both the agent AND the ns-init host pid are gone.
+    deadline = time.monotonic() + 20.0
+    while time.monotonic() < deadline and (
+        _proc_alive(tracked.agent.pid) or _proc_alive(tracked.inner_pid)
+    ):
+        await asyncio.sleep(0.25)
+    assert not _proc_alive(tracked.agent.pid), "agent pid survived destroy (lifecycle)"
+    assert not _proc_alive(tracked.inner_pid), (
+        "inner namespace init survived destroy — leaked sandbox (tmpfs + bind "
+        "held for a full hour; every destroy leaked one namespace process)"
+    )
+
+
+def _proc_alive(pid: int | None) -> bool:
+    if pid is None:
+        return False
+    try:
+        with open(f"/proc/{pid}/stat") as fh:
+            # state is the first field after comm: Z (zombie) counts as dead
+            # for our purpose — it holds no namespace and is reaped next.
+            return fh.read().rsplit(") ", 1)[1].split()[0] != "Z"
+    except OSError:
+        return False
+
+
+async def _await_events(
+    store: SQLiteTraceStore, learner: str, task: str, *, minimum: int, timeout_s: float = 20.0
+) -> list:
+    """Poll the store until `minimum` events arrive (agent streams async)."""
+    deadline = time.monotonic() + timeout_s
+    while time.monotonic() < deadline:
+        events = store.get_trace(learner, task)
+        if len(events) >= minimum:
+            return events
+        await asyncio.sleep(0.25)
+    events = store.get_trace(learner, task)
+    pytest.fail(
+        f"only {len(events)}/{minimum} telemetry events arrived within {timeout_s}s: "
+        f"{[(e.seq, e.kind) for e in events]}"
+    )

apps/ai-service/tests/telemetry/test_durability.py

@@ -0,0 +1,239 @@
+"""Dropped-connection durability probe (Task 2-4-01, REQ-3-003).
+
+At-least-once delivery + server-side (learner,task,seq) dedup = every event
+stored exactly once, in order — proven against a REAL uvicorn ingest and a
+REAL namespace sandbox, with the connection severed mid-stream by a killable
+TCP proxy (deterministic "network down" window).
+
+Probe-guarded: skips (not fails) on hosts without user namespaces.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import contextlib
+import socket
+import struct
+import time
+from pathlib import Path
+
+import pytest
+import uvicorn
+
+from ai_service.config import Settings
+from ai_service.main import create_app
+from ai_service.sandbox import SandboxManager
+from ai_service.sandbox.unshare_backend import UnshareBackend
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+
+from ..sandbox.test_isolation import USERSNS_AVAILABLE
+
+
+def _userns_probe() -> None:
+    if not USERSNS_AVAILABLE:
+        pytest.skip("user namespaces unavailable on this host (probe)")
+
+
+def _free_port() -> int:
+    with socket.socket() as s:
+        s.bind(("127.0.0.1", 0))
+        return s.getsockname()[1]
+
+
+class KillableProxy:
+    """TCP forwarder whose data path dies on ``kill()`` — the listener stays.
+
+    Accepts are preserved so the agent's reconnects fail fast (connection
+    reset) instead of hanging on a black-holed socket, keeping the outage
+    window deterministic. ``revive()`` restores forwarding.
+    """
+
+    def __init__(self, target_port: int) -> None:
+        self.target_port = target_port
+        self.listen_port = _free_port()
+        self._listener: socket.socket | None = None
+        self._killed = False
+        self._conns: list[socket.socket] = []
+
+    async def start(self) -> None:
+        self._listener = socket.socket()
+        self._listener.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+        self._listener.bind(("127.0.0.1", self.listen_port))
+        self._listener.listen(16)
+        self._listener.setblocking(False)
+        loop = asyncio.get_running_loop()
+        loop.create_task(self._accept_loop())
+
+    async def _accept_loop(self) -> None:
+        loop = asyncio.get_running_loop()
+        assert self._listener is not None
+        while True:
+            try:
+                client, _ = await loop.sock_accept(self._listener)
+            except OSError:
+                return  # listener closed
+            client.setblocking(False)  # accepted sockets default to blocking
+            if self._killed:
+                with contextlib.suppress(OSError):
+                    # RST on accept: reconnects fail fast while "down"
+                    client.setsockopt(
+                        socket.SOL_SOCKET, socket.SO_LINGER,
+                        struct.pack("<ii", 1, 0),
+                    )
+                    client.close()
+                continue
+            upstream = socket.socket()
+            upstream.setblocking(False)
+            try:
+                await loop.sock_connect(upstream, ("127.0.0.1", self.target_port))
+            except OSError:
+                with contextlib.suppress(OSError):
+                    client.close()
+                continue
+            self._conns.extend((client, upstream))
+            loop.create_task(self._pump(client, upstream))
+            loop.create_task(self._pump(upstream, client))
+
+    async def _pump(self, src: socket.socket, dst: socket.socket) -> None:
+        """Forward until EOF, error, or the killed flag is observed.
+
+        Only ``dst`` is closed here — the reverse-direction pump owns ``src``
+        (closing the peer's socket from this task causes EBADF cascades).
+        """
+        loop = asyncio.get_running_loop()
+        try:
+            while True:
+                data = await loop.sock_recv(src, 65536)
+                if not data:
+                    return
+                if self._killed:
+                    return  # outage: drop the relay, peer sees EOF on dst close
+                await loop.sock_sendall(dst, data)
+        except OSError:
+            return
+        finally:
+            with contextlib.suppress(OSError):
+                dst.close()
+
+    def kill(self) -> None:
+        """Sever the data path — flag only.
+
+        The pumps and the accept loop observe ``_killed`` themselves and close
+        THEIR OWN sockets from inside the event loop. Closing sockets that the
+        loop is currently awaiting from the outside deadlocks the loop's
+        child-watcher in this environment (observed: post-kill subprocess
+        spawns hung forever) — so no external socket surgery, ever.
+        """
+
+    def revive(self) -> None:
+        self._killed = False
+
+
+async def _await_events(
+    store: SQLiteTraceStore, learner: str, task: str, *, minimum: int, timeout_s: float = 30.0
+) -> list:
+    deadline = time.monotonic() + timeout_s
+    while time.monotonic() < deadline:
+        events = store.get_trace(learner, task)
+        if len(events) >= minimum:
+            return events
+        await asyncio.sleep(0.25)
+    return store.get_trace(learner, task)
+
+
+@pytest.mark.asyncio
+async def test_disconnect_reconnect_loses_nothing(tmp_path: Path) -> None:
+    """Sever the agent's WS mid-stream; every event lands exactly once, ordered."""
+    _userns_probe()
+
+    learner = "durability-learner"
+    task = "task-durability"
+
+    store = SQLiteTraceStore(db_path=tmp_path / "durability.db")
+    app = create_app()
+    app.state.trace_store = store
+    app.state.trace_integrity = TraceIntegrityMap()
+
+    server_port = _free_port()
+    server = uvicorn.Server(
+        uvicorn.Config(app, host="127.0.0.1", port=server_port, log_level="warning")
+    )
+    serve_task = asyncio.get_running_loop().create_task(server.serve())
+
+    proxy = KillableProxy(target_port=server_port)
+    await proxy.start()
+
+    manager = SandboxManager(backend=UnshareBackend(), settings=Settings())
+    app.state.sandbox_manager = manager
+
+    def capture_env(sandbox_id: str, learner_id: str, task_id: str) -> dict[str, str]:
+        return {
+            "NC_LEARNER_ID": learner_id,
+            "NC_TASK_ID": task_id,
+            "NC_SANDBOX_ID": sandbox_id,
+            # Agent dials the PROXY; the proxy forwards to the real server.
+            "NC_INGEST_URL": (
+                f"ws://127.0.0.1:{proxy.listen_port}/v1/telemetry/ingest"
+                f"?learner_id={learner_id}&task_id={task_id}&sandbox_id={sandbox_id}"
+            ),
+            # Fast reconnect so the probe stays bounded.
+            "NC_BACKOFF_BASE_S": "0.1",
+            "NC_BACKOFF_MAX_S": "0.5",
+        }
+
+    manager._capture_env = capture_env  # noqa: SLF001 - test seam
+
+    try:
+        for _ in range(100):
+            if server.started:
+                break
+            await asyncio.sleep(0.1)
+        assert server.started, "uvicorn did not start"
+
+        handle = await manager.create(learner, task_id=task)
+        try:
+            live = manager._handles[handle.id]  # noqa: SLF001 - test seam
+            backend: UnshareBackend = manager._backend  # noqa: SLF001 - test seam
+
+            # Phase A — connected: a write streams through.
+            result = await backend.exec(live, ["sh", "-c", "echo one > a.txt"])
+            assert result.returncode == 0, result.stderr
+            events = await _await_events(store, learner, task, minimum=1)
+            assert events, "no events arrived before the outage"
+
+            # Phase B — network down: sever mid-stream, keep generating.
+            proxy.kill()
+            for n in ("two", "three", "four"):
+                result = await backend.exec(live, ["sh", "-c", f"echo {n} > {n}.txt"])
+                assert result.returncode == 0, result.stderr
+                await asyncio.sleep(0.3)
+
+            # Phase C — revive: the agent reconnects (fast backoff) and
+            # flushes the spool. Server dedups on (learner,task,seq).
+            proxy.revive()
+            await asyncio.sleep(1.5)
+            for n in ("five", "six"):
+                result = await backend.exec(live, ["sh", "-c", f"echo {n} > {n}.txt"])
+                assert result.returncode == 0, result.stderr
+
+            events = await _await_events(
+                store, learner, task, minimum=6, timeout_s=30.0
+            )
+            seqs = [e.seq for e in events]
+            assert seqs == sorted(seqs), f"out of order after reconnect: {seqs}"
+            assert len(set(seqs)) == len(seqs), f"duplicates stored: {seqs}"
+            assert len(events) >= 6, f"lost events across the outage: {seqs}"
+            # Exactly-once storage despite at-least-once delivery: every spool
+            # flush may re-send the replay-margin line; dedup collapses it.
+            stored_files = {
+                e.payload.get("path") for e in events if e.kind == "file_diff"
+            }
+            assert stored_files, "file_diff events missing from the stored trace"
+        finally:
+            await manager.destroy(handle.id)
+    finally:
+        server.should_exit = True
+        with contextlib.suppress(Exception):
+            await asyncio.wait_for(serve_task, timeout=10.0)
+        store.close()

apps/ai-service/tests/telemetry/test_models.py

@@ -0,0 +1,126 @@
+"""TelemetryEvent model contract tests (REQ-3-003, D-027).
+
+Pure validation tests — no store, no network, no fixtures beyond a
+canonical kwargs builder.
+"""
+
+import json
+from datetime import UTC, datetime
+
+import pytest
+from pydantic import ValidationError
+
+from ai_service.telemetry import EventKind, TelemetryEvent, TraceSpan
+
+
+def make_event_kwargs(**overrides) -> dict:
+    base = {
+        "learner_id": "learner-1",
+        "task_id": "task-1",
+        "seq": 0,
+        "kind": "command",
+        "payload": {"argv": ["pytest"], "cwd": "/workspace"},
+        "ts": datetime(2026, 9, 11, tzinfo=UTC),
+        "sandbox_id": "sbx-1",
+    }
+    base.update(overrides)
+    return base
+
+
+def make_event(**overrides) -> TelemetryEvent:
+    return TelemetryEvent(**make_event_kwargs(**overrides))
+
+
+def test_valid_event_constructs_with_all_kinds():
+    for kind in (
+        "command",
+        "file_diff",
+        "run_result",
+        "test_result",
+        "activity",
+        "stdin",
+        "stdout",
+    ):
+        event = make_event(kind=kind)
+        assert event.kind == kind
+
+
+def test_zero_seq_accepted():
+    assert make_event(seq=0).seq == 0
+
+
+def test_negative_seq_rejected():
+    # @validates hooks raise ValueError (not pydantic ValidationError) at
+    # construction — same for kind and empty-id checks below.
+    with pytest.raises(ValueError, match="seq must be >= 0"):
+        make_event(seq=-1)
+
+
+def test_bad_kind_rejected():
+    with pytest.raises(ValueError, match="unknown event kind"):
+        make_event(kind="keystroke")  # not in the EventKind literal
+
+
+def test_empty_learner_id_rejected():
+    with pytest.raises(ValueError, match="non-empty identifier"):
+        make_event(learner_id="")
+
+
+def test_empty_task_id_rejected():
+    with pytest.raises(ValueError, match="non-empty identifier"):
+        make_event(task_id="")
+
+
+def test_payload_json_roundtrip():
+    payload = {
+        "command": "pytest -q",
+        "exit_code": 1,
+        "durations": [0.12, 3.4],
+        "nested": {"passed": 7, "failed": 2},
+        "unicode": "héllo",
+    }
+    event = make_event(payload=payload)
+    assert event.payload == payload
+    # JSON-serializable payloads survive a full dumps/loads roundtrip.
+    assert json.loads(json.dumps(event.payload)) == payload
+
+
+def test_event_json_roundtrip():
+    event = make_event(payload={"stdout": "ok", "n": 3})
+    restored = TelemetryEvent.model_validate_json(event.model_dump_json())
+    # pydantic leaves a JSON-str ts as str; compare field-by-field instead of
+    # dataclass equality, which distinguishes 'Z' string vs parsed datetime.
+    assert restored.learner_id == event.learner_id
+    assert restored.task_id == event.task_id
+    assert restored.seq == event.seq
+    assert restored.kind == event.kind
+    assert restored.payload == event.payload
+    assert restored.sandbox_id == event.sandbox_id
+    # SQLModel leaves a JSON-serialized ts as its str form; parse to compare.
+    restored_ts = datetime.fromisoformat(str(restored.ts).replace("Z", "+00:00"))
+    assert restored_ts == event.ts
+
+
+def test_tracespan_holds_ordered_events():
+    events = tuple(make_event(seq=seq, kind="file_diff") for seq in range(3))
+    span = TraceSpan(learner_id="learner-1", task_id="task-1", events=events)
+    assert [e.seq for e in span.events] == [0, 1, 2]
+    assert span.latest_seq == 2
+
+
+def test_tracespan_empty_has_no_latest_seq():
+    span = TraceSpan(learner_id="learner-1", task_id="task-1")
+    assert span.events == ()
+    assert span.latest_seq == -1
+
+
+def test_tracespan_rejects_empty_ids():
+    with pytest.raises(ValidationError):
+        TraceSpan(learner_id="", task_id="task-1")
+    with pytest.raises(ValidationError):
+        TraceSpan(learner_id="learner-1", task_id="")
+
+
+def test_event_kind_type_is_exported_literal():
+    # The kind discriminator is part of the public module surface.
+    assert "command" in EventKind.__args__

apps/ai-service/tests/telemetry/test_store.py

@@ -0,0 +1,206 @@
+"""SQLiteTraceStore tests (REQ-3-003, D-027).
+
+Each test gets its own tmp-path SQLite file — no shared disk state. Covers:
+    - append / ordered get (events may arrive out of order; reads come back
+      ordered by seq)
+    - dedup on retry: the same (learner_id, task_id, seq) appended twice is
+      stored exactly once (at-least-once ingest contract)
+    - retry dedup must not overwrite the originally stored row
+    - gap detection, latest_seq, list_tasks
+    - WAL + synchronous=NORMAL pragmas actually applied to the DB file
+    - concurrent writer + reader against the same DB file (a-3 smoke test)
+"""
+
+import concurrent.futures
+import threading
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+from typing import Any
+
+import pytest
+import sqlalchemy as sa
+
+from ai_service.telemetry.models import EventKind, TelemetryEvent
+from ai_service.telemetry.store import SQLiteTraceStore
+
+_BASE_TS = datetime(2026, 9, 11, 12, 0, 0, tzinfo=UTC)
+
+
+def make_event(
+    seq: int,
+    learner_id: str = "learner-1",
+    task_id: str = "task-1",
+    kind: EventKind = "command",
+    payload: dict[str, Any] | None = None,
+    sandbox_id: str = "sbx-1",
+) -> TelemetryEvent:
+    return TelemetryEvent(
+        learner_id=learner_id,
+        task_id=task_id,
+        seq=seq,
+        kind=kind,
+        payload=payload if payload is not None else {"seq": seq},
+        ts=_BASE_TS + timedelta(seconds=seq),
+        sandbox_id=sandbox_id,
+    )
+
+
+@pytest.fixture
+def store(tmp_path: Path):
+    s = SQLiteTraceStore(db_path=tmp_path / "telemetry.db")
+    yield s
+    s.close()
+
+
+def test_append_and_get_trace_orders_by_seq(store: SQLiteTraceStore) -> None:
+    # Append out of order; reads must come back ordered by seq.
+    store.append(make_event(2, kind="stdout"))
+    store.append(make_event(0, kind="stdin"))
+    store.append(make_event(1, kind="run_result"))
+
+    trace = store.get_trace("learner-1", "task-1")
+    assert [e.seq for e in trace] == [0, 1, 2]
+    assert [e.kind for e in trace] == ["stdin", "run_result", "stdout"]
+    assert all(e.learner_id == "learner-1" for e in trace)
+    assert all(e.task_id == "task-1" for e in trace)
+
+
+def test_get_trace_round_trips_fields(store: SQLiteTraceStore) -> None:
+    event = make_event(0, payload={"file": "a.py", "nested": {"ok": True}})
+    store.append(event)
+
+    (row,) = store.get_trace("learner-1", "task-1")
+    assert row.learner_id == event.learner_id
+    assert row.task_id == event.task_id
+    assert row.seq == 0
+    assert row.kind == "command"
+    assert row.payload == {"file": "a.py", "nested": {"ok": True}}
+    assert row.ts == _BASE_TS
+    assert row.sandbox_id == "sbx-1"
+
+
+def test_get_trace_is_scoped_to_the_task_pair(store: SQLiteTraceStore) -> None:
+    store.append(make_event(0, learner_id="learner-1", task_id="task-1"))
+    store.append(make_event(0, learner_id="learner-1", task_id="task-2"))
+    store.append(make_event(0, learner_id="learner-2", task_id="task-1"))
+
+    assert [e.seq for e in store.get_trace("learner-1", "task-1")] == [0]
+    assert len(store.get_trace("learner-1", "task-2")) == 1
+    assert len(store.get_trace("learner-2", "task-1")) == 1
+    assert store.get_trace("learner-1", "task-missing") == []
+
+
+def test_append_is_idempotent_on_retry(store: SQLiteTraceStore) -> None:
+    # At-least-once ingest re-delivers the same event (same idempotency key).
+    # It must be stored exactly once and the retry must be a no-op success.
+    original = make_event(0, kind="command", payload={"attempt": 1})
+    store.append(original)
+    store.append(original)
+    # A distinct-but-conflicting retry (same key, different body) is also
+    # deduped — the first stored row wins, no overwrite.
+    retry = make_event(0, kind="file_diff", payload={"attempt": 2})
+    store.append(retry)
+
+    trace = store.get_trace("learner-1", "task-1")
+    assert len(trace) == 1
+    assert trace[0].kind == "command"
+    assert trace[0].payload == {"attempt": 1}
+    assert store.latest_seq("learner-1", "task-1") == 0
+
+
+def test_gaps_reports_missing_seqs(store: SQLiteTraceStore) -> None:
+    for seq in (0, 2, 3, 7):
+        store.append(make_event(seq))
+
+    assert store.gaps("learner-1", "task-1") == [1, 4, 5, 6]
+    # Gaps are per-trace: an empty trace has no gaps at all.
+    assert store.gaps("learner-1", "task-unknown") == []
+
+
+def test_latest_seq(store: SQLiteTraceStore) -> None:
+    assert store.latest_seq("learner-1", "task-1") == -1
+
+    store.append(make_event(0))
+    assert store.latest_seq("learner-1", "task-1") == 0
+
+    store.append(make_event(5))  # gaps do not move latest_seq
+    assert store.latest_seq("learner-1", "task-1") == 5
+    # Scoped to the trace pair.
+    assert store.latest_seq("learner-1", "task-2") == -1
+
+
+def test_count_is_durable_row_count_not_latest_seq(store: SQLiteTraceStore) -> None:
+    """count() backs the ingest flood cap (P7): it must reflect stored ROWS
+    (a skipped-ahead seq must not burn un-sent budget) and stay O(1)-ish
+    (COUNT(*), never materialize the trace per append)."""
+    assert store.count("learner-1", "task-1") == 0
+    store.append(make_event(0))
+    store.append(make_event(2))  # skipped 1 — count is rows, not latest+1
+    assert store.count("learner-1", "task-1") == 2
+    # Dedup retries do not inflate the count (at-least-once contract).
+    store.append(make_event(2))
+    assert store.count("learner-1", "task-1") == 2
+    # Scoped to the trace pair.
+    assert store.count("learner-1", "task-2") == 0
+
+
+def test_list_tasks(store: SQLiteTraceStore) -> None:
+    assert store.list_tasks("learner-1") == []
+
+    store.append(make_event(0, task_id="task-b"))
+    store.append(make_event(1, task_id="task-b"))
+    store.append(make_event(0, task_id="task-a"))
+
+    assert store.list_tasks("learner-1") == ["task-a", "task-b"]
+    # Scoped per learner.
+    assert store.list_tasks("learner-2") == []
+
+
+def test_pragmas_are_applied(store: SQLiteTraceStore) -> None:
+    # Pragmas are per-connection; query through the store's engine so the
+    # connect hook (not a default sqlite3 connection) is what we inspect.
+    with store._engine.connect() as conn:
+        (journal_mode,) = conn.execute(sa.text("PRAGMA journal_mode")).one()
+        (synchronous,) = conn.execute(sa.text("PRAGMA synchronous")).one()
+    assert journal_mode == "wal"
+    # synchronous=NORMAL is 1 in SQLite's pragma numbering.
+    assert synchronous == 1
+
+
+def test_concurrent_writer_and_reader_no_database_is_locked(tmp_path: Path) -> None:
+    """One thread appends while another reads in a tight loop (a-3).
+
+    Without WAL + busy_timeout this pattern reliably produces
+    `OperationalError: database is locked` on SQLite. The assertion is that
+    every reader call completes and the final trace is complete.
+    """
+    db_path = tmp_path / "telemetry.db"
+    n_events = 60
+    stop_writing = threading.Event()
+
+    writer = SQLiteTraceStore(db_path=db_path)
+    reader = SQLiteTraceStore(db_path=db_path)
+    try:
+
+        def write_events() -> None:
+            for seq in range(n_events):
+                writer.append(make_event(seq))
+            stop_writing.set()
+
+        def read_trace() -> None:
+            while not stop_writing.is_set():
+                reader.get_trace("learner-1", "task-1")
+            # Final read after the writer is done.
+            assert len(reader.get_trace("learner-1", "task-1")) == n_events
+
+        with concurrent.futures.ThreadPoolExecutor(max_workers=2) as pool:
+            futures = [pool.submit(write_events), pool.submit(read_trace)]
+            for future in futures:
+                future.result(timeout=30)
+
+        assert [e.seq for e in writer.get_trace("learner-1", "task-1")] == list(
+            range(n_events)
+        )
+    finally:
+        reader.close()
+        writer.close()

apps/ai-service/tests/test_corpus_dormancy.py

@@ -0,0 +1,79 @@
+"""Corpus dormancy verification (Task 6-1-04, REQ-3-007).
+
+The v0.2 mock engine inputs (`corpus/telemetry.py`, `corpus/artifacts.py`)
+must have ZERO production importers after the v0.3 re-grounding: Lab/
+Assessor/Proctor run on real engine inputs with no mock fallback in the
+learner path. The files stay on disk (Phase-3 calibration history) but are
+not imported by any production module. `learner_context` remains ACTIVE
+(agents still need learner context). Test-only references (e.g.
+`corpus/trace_fixtures.py` in grading calibration tests) are allowed.
+"""
+
+from __future__ import annotations
+
+import ast
+from pathlib import Path
+
+REPO = Path(__file__).parents[1]
+
+#: Production trees whose imports of dormant corpus modules are forbidden.
+PRODUCTION_PATHS = [
+    REPO / "ai_service" / "agents",
+    REPO / "ai_service" / "api",
+    REPO / "ai_service" / "grading",
+    REPO / "ai_service" / "telemetry",
+    REPO / "ai_service" / "variants",
+    REPO / "ai_service" / "voice",
+    REPO / "ai_service" / "sandbox",
+    REPO / "ai_service" / "llm",
+    REPO / "ai_service" / "main.py",
+]
+
+DORMANT_MODULES = ("corpus.telemetry", "corpus.artifacts")
+
+
+def _module_targets(node: ast.AST, *, level: int, module: str | None) -> set[str]:
+    """Resolve relative + absolute import targets to dotted ai_service paths."""
+    targets: set[str] = set()
+    if module and ("corpus" in module):
+        targets.add(module)
+    return targets
+
+
+def test_no_production_imports_of_dormant_corpus() -> None:
+    violators: list[str] = []
+    for path in PRODUCTION_PATHS:
+        files = [path] if path.suffix == ".py" else sorted(path.rglob("*.py"))
+        for py in files:
+            tree = ast.parse(py.read_text())
+            for node in ast.walk(tree):
+                if isinstance(node, ast.ImportFrom):
+                    module = node.module or ""
+                    level = node.level
+                    if level:  # relative: resolve against corpus
+                        if module.endswith("telemetry") and "corpus" in module:
+                            violators.append(f"{py}: {module}")
+                        if module.endswith("artifacts") and "corpus" in module:
+                            violators.append(f"{py}: {module}")
+                        # bare `from . import telemetry` inside corpus/ itself is fine
+                    else:
+                        for target in _module_targets(node, level=level, module=module):
+                            violators.append(f"{py}: {target}")
+                elif isinstance(node, ast.Import):
+                    for alias in node.names:
+                        if any(alias.name.startswith(m) for m in DORMANT_MODULES):
+                            violators.append(f"{py}: {alias.name}")
+    assert not violators, f"dormant corpus imports in production: {violators}"
+
+
+def test_learner_context_stays_active() -> None:
+    """Learner context corpus is NOT dormant — agents still use it."""
+    agents_lab = (REPO / "ai_service" / "agents" / "lab.py").read_text()
+    assert "corpus.learner_context" in agents_lab
+    (REPO / "ai_service" / "corpus" / "learner_context.py").exists()
+
+
+def test_dormancy_headers_present() -> None:
+    for fname in ("telemetry.py", "artifacts.py"):
+        src = (REPO / "ai_service" / "corpus" / fname).read_text()
+        assert "DORMANT" in src, f"{fname} missing dormancy header"

apps/ai-service/tests/variants/__init__.py

@@ -0,0 +1 @@
+"""Variant task generation tests (REQ-3-005)."""

apps/ai-service/tests/variants/test_generator.py

@@ -0,0 +1,179 @@
+"""Variant generator tests (Task 4-2-01, REQ-3-005) — D-029 + a-5 binding."""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime, timedelta
+
+import pytest
+
+from ai_service.grading.features import compute_digest
+from ai_service.llm.mock import MockProvider
+from ai_service.telemetry.models import TelemetryEvent
+from ai_service.variants.generator import (
+    MILESTONE,
+    VariantGenerator,
+    derive_seed,
+    derive_task_id,
+)
+from ai_service.variants.store import SQLiteVariantStore, VariantRecord
+from ai_service.variants.templates import TEMPLATES, get_template
+
+
+class ScriptedRenderProvider(MockProvider):
+    """Deterministic render: the statement embeds the params (distinct per draw)."""
+
+    def __init__(self) -> None:
+        super().__init__()
+        self.calls = 0
+
+    async def chat(self, messages, model, response_format=None):  # noqa: ANN001
+        self.calls += 1
+        import json
+
+        user = next(m.content for m in reversed(messages) if m.role == "user")
+        # Distinct per distinct params: hash the seeded slot lines.
+        fingerprint = abs(hash(user)) % 10_000
+        return json.dumps({"statement": f"Scripted variant #{fingerprint} — build it."})
+
+
+class FailingRenderProvider(MockProvider):
+    """Always fails D-020 validation -> deterministic fallback path."""
+
+    def __init__(self) -> None:
+        super().__init__()
+        self.calls = 0
+
+    async def chat(self, messages, model, response_format=None):  # noqa: ANN001
+        self.calls += 1
+        return "this is not json at all"
+
+
+@pytest.fixture()
+def store(tmp_path):  # noqa: ANN001
+    s = SQLiteVariantStore(db_path=tmp_path / "variants.db")
+    yield s
+    s.close()
+
+
+async def test_two_learners_distinct_statements(store) -> None:  # noqa: ANN001
+    provider = ScriptedRenderProvider()
+    gen = VariantGenerator(store, provider, model="mock")
+    a = await gen.generate("learner-a", "tpl-llm-judge")
+    b = await gen.generate("learner-b", "tpl-llm-judge")
+    assert a.statement != b.statement
+    assert a.seed != b.seed
+    assert a.task_id != b.task_id
+
+
+async def test_same_learner_is_cached_no_second_llm_call(store) -> None:  # noqa: ANN001
+    provider = ScriptedRenderProvider()
+    gen = VariantGenerator(store, provider, model="mock")
+    first = await gen.generate("learner-a", "tpl-llm-judge")
+    calls_after_first = provider.calls
+    second = await gen.generate("learner-a", "tpl-llm-judge")
+    assert first == second  # identical stored variant (D-029 reproducible)
+    assert provider.calls == calls_after_first  # cache hit: NO LLM call
+
+
+async def test_seed_derivation_reproducible() -> None:
+    s1 = derive_seed("tpl-llm-judge", "learner-a", MILESTONE)
+    s2 = derive_seed("tpl-llm-judge", "learner-a", MILESTONE)
+    assert s1 == s2
+    assert derive_seed("tpl-llm-judge", "learner-b", MILESTONE) != s1
+    assert derive_task_id(s1).startswith("task-")
+    assert len(derive_task_id(s1)) == len("task-") + 16
+
+
+async def test_params_are_schema_valid(store) -> None:  # noqa: ANN001
+    gen = VariantGenerator(store, ScriptedRenderProvider(), model="mock")
+    record = await gen.generate("learner-a", "tpl-guardrail-schema")
+    template = get_template("tpl-guardrail-schema")
+    for slot in template.slots:  # type: ignore[union-attr]
+        value = record.params[slot.name]
+        assert slot.validate_value(value), f"slot {slot.name} drew invalid {value!r}"
+
+
+async def test_llm_failure_falls_back_deterministically(store) -> None:  # noqa: ANN001
+    provider = FailingRenderProvider()
+    gen = VariantGenerator(store, provider, model="mock")
+    record = await gen.generate("learner-a", "tpl-rag-chunker")
+    template = get_template("tpl-rag-chunker")
+    expected = template.render({k: v for k, v in record.params.items()})  # type: ignore
+    assert record.statement == expected  # skeleton render, seed-auditable
+    assert provider.calls == 2  # D-020 bounded retry, then fallback
+
+
+async def test_unknown_template_raises(store) -> None:  # noqa: ANN001
+    gen = VariantGenerator(store, ScriptedRenderProvider(), model="mock")
+    with pytest.raises(ValueError, match="no task template"):
+        await gen.generate("learner-a", "tpl-does-not-exist")
+
+
+def test_fairness_envelope_same_bar_per_template() -> None:
+    """a-5 (BINDING): every legal variant of one template fits the anchors.
+
+    For 10 different learners: draw the seeded params, then synthesize a
+    trace whose edit count is sampled INSIDE the template's anchor band and
+    whose test runs meet the anchor minimum — the resulting digests must
+    all sit within the template's expected feature envelope. That is the
+    testable form of "same bar": no slot draw can push a variant outside
+    the effort band the grader context assumes.
+    """
+    import random
+
+    for template in TEMPLATES.values():
+        anchors = template.rubric_anchors
+        lo_edits, hi_edits = anchors.expected_edit_count_band
+        t0 = datetime(2026, 9, 12, tzinfo=UTC)
+        for i in range(10):
+            params = template.sample_params(seed=10_000 + i)
+            rng = random.Random(i)
+            n_edits = rng.randint(lo_edits, hi_edits)
+            events = [
+                TelemetryEvent(
+                    learner_id=f"fair-learner-{i}",
+                    task_id=f"fair-task-{i}",
+                    seq=n,
+                    kind="file_diff",
+                    payload={"path": f"f{n}.py"},
+                    ts=t0 + timedelta(seconds=n * 10),
+                    sandbox_id="sbx-fair",
+                )
+                for n in range(n_edits)
+            ]
+            # Meet the anchor's minimum test-run expectation.
+            for t in range(anchors.expected_min_test_runs):
+                events.append(
+                    TelemetryEvent(
+                        learner_id=f"fair-learner-{i}",
+                        task_id=f"fair-task-{i}",
+                        seq=len(events),
+                        kind="test_result",
+                        payload={"passed": t == anchors.expected_min_test_runs - 1},
+                        ts=t0 + timedelta(seconds=(n_edits + t) * 10),
+                        sandbox_id="sbx-fair",
+                    )
+                )
+            digest = compute_digest(events)
+            assert lo_edits <= digest.edit_count <= hi_edits
+            assert digest.test_pass_count + digest.test_fail_count >= (
+                anchors.expected_min_test_runs
+            )
+            # Slot values never appear in the digest (no scenario leakage into
+            # grading features — difficulty stays scenario-independent).
+            digest_json = digest.model_dump_json()
+            for value in params.values():
+                assert str(value) not in digest_json or isinstance(value, int)
+
+
+async def test_variant_record_roundtrips_through_store(store) -> None:  # noqa: ANN001
+    gen = VariantGenerator(store, ScriptedRenderProvider(), model="mock")
+    record = await gen.generate("learner-a", "tpl-llm-judge")
+    fetched = store.get("learner-a", "tpl-llm-judge")
+    assert fetched is not None
+    assert fetched.statement == record.statement
+    assert fetched.seed == record.seed
+    by_task = store.get_by_task(record.task_id)
+    assert by_task is not None
+    assert by_task.learner_id == "learner-a"
+    assert isinstance(record, VariantRecord)

apps/ai-service/tests/variants/test_store.py

@@ -0,0 +1,358 @@
+"""SQLiteVariantStore tests (REQ-3-005, D-027).
+
+Each test gets its own tmp-path SQLite file — no shared disk state. Covers:
+    - save / get roundtrip by (learner_id, template_id) AND by task_id
+      (all fields survive, including nested JSON params, starter_files
+      filename->content map, and the tz-aware created_at contract)
+    - insert-only: a second save for the same (learner_id, template_id)
+      raises IntegrityError (first-wins, documented choice); the original
+      row is untouched — NOT upsert, NOT swallowed
+    - unique task_id: a second variant claiming an existing trace key is
+      rejected even under a different (learner, template) pair
+    - list_for_learner / list_by_template scoped + chronological + auditable
+      (seed + params readable back — the proctoring cross-check path)
+    - unknown learner / template / task -> None / empty lists
+    - scoping: rows for other learners/templates never leak
+    - rows are detached: usable after the store is closed
+    - WAL + synchronous=NORMAL pragmas actually applied to the DB file
+    - concurrent writer + reader against the same DB file (a-3 smoke test)
+"""
+
+import concurrent.futures
+import threading
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+from typing import Any
+
+import pytest
+import sqlalchemy as sa
+
+from ai_service.variants.store import SQLiteVariantStore, VariantRecord
+
+_BASE_TS = datetime(2026, 9, 12, 12, 0, 0, tzinfo=UTC)
+
+
+def make_variant(
+    task_id: str = "task-1",
+    learner_id: str = "learner-1",
+    template_id: str = "template-1",
+    seed: str = "seed-a1b2c3",
+    params: dict[str, Any] | None = None,
+    statement: str | None = None,
+    starter_files: dict[str, Any] | None = None,
+    created_at: datetime | None = None,
+) -> VariantRecord:
+    """Canonical kwargs builder — tests override only what they assert on."""
+    return VariantRecord(
+        learner_id=learner_id,
+        template_id=template_id,
+        task_id=task_id,
+        seed=seed,
+        params=params
+        if params is not None
+        else {
+            "scenario": "cache invalidation",
+            "constraints": ["no external deps", "streaming"],
+            "data_shape": {"rows": 10_000, "columns": ["ts", "event", "user"]},
+        },
+        statement=statement
+        if statement is not None
+        else (
+            "Implement a cache layer for the event stream service that "
+            "survives restarts without losing buffered rows, learner "
+            f"{learner_id} edition."
+        ),
+        starter_files=starter_files
+        if starter_files is not None
+        else {
+            "src/stream_cache.py": "class StreamCache:\n    pass\n",
+            "tests/test_stream_cache.py": "def test_roundtrip():\n    pass\n",
+        },
+        created_at=created_at if created_at is not None else _BASE_TS,
+    )
+
+
+@pytest.fixture
+def store(tmp_path: Path) -> SQLiteVariantStore:
+    s = SQLiteVariantStore(db_path=tmp_path / "variants.db")
+    yield s
+    s.close()
+
+
+def test_save_and_get_roundtrip(store: SQLiteVariantStore) -> None:
+    variant = make_variant()
+    store.save(variant)
+
+    # Point lookup by variant identity (learner_id, template_id).
+    fetched = store.get("learner-1", "template-1")
+    assert fetched is not None
+    assert fetched.learner_id == "learner-1"
+    assert fetched.template_id == "template-1"
+    assert fetched.task_id == "task-1"
+    assert fetched.seed == "seed-a1b2c3"
+    assert fetched.params == variant.params
+    assert fetched.statement == variant.statement
+    assert fetched.starter_files == variant.starter_files
+    assert fetched.created_at == _BASE_TS
+    assert fetched.created_at.tzinfo is UTC  # tz-normalized on read
+
+    # Same row via the grading/telemetry trace key — the join path a
+    # grade or proctor check uses with only a task_id in hand.
+    by_task = store.get_by_task("task-1")
+    assert by_task is not None
+    assert by_task.learner_id == "learner-1"
+    assert by_task.template_id == "template-1"
+    assert by_task.seed == "seed-a1b2c3"
+    assert by_task.statement == variant.statement
+    assert by_task.starter_files == variant.starter_files
+    assert by_task.params == variant.params
+    assert by_task.created_at.tzinfo is UTC
+
+
+def test_get_unknown_returns_none(store: SQLiteVariantStore) -> None:
+    store.save(make_variant())
+
+    assert store.get("learner-1", "template-missing") is None
+    assert store.get("learner-missing", "template-1") is None
+    assert store.get("nobody", "nothing") is None
+    assert store.get_by_task("task-missing") is None
+
+
+def test_duplicate_learner_template_pair_raises_integrity_error(
+    store: SQLiteVariantStore,
+) -> None:
+    """THE contract of this store (insert-only, first wins).
+
+    The first generated variant is authoritative (D-029 reproducibility:
+    the seed re-derives the same variant); a duplicate save is a
+    programming error or lost race, not a normal flow — the generator's
+    cache path serves `get` instead. So the IntegrityError surfaces to
+    the caller and the stored row is left untouched.
+    """
+    first = make_variant(statement="first authoritative statement")
+    store.save(first)
+
+    second = make_variant(
+        task_id="task-2",  # distinct task key; the PAIR collides
+        seed="seed-999",
+        statement="an impostor statement",
+        created_at=_BASE_TS + timedelta(hours=1),
+    )
+    with pytest.raises(sa.exc.IntegrityError):
+        store.save(second)
+
+    # First save survived intact — nothing was overwritten.
+    fetched = store.get("learner-1", "template-1")
+    assert fetched is not None
+    assert fetched.task_id == "task-1"
+    assert fetched.statement == "first authoritative statement"
+    assert fetched.seed == "seed-a1b2c3"
+    assert fetched.created_at == _BASE_TS
+
+    # The rejected save left no row behind under its task_id either.
+    assert store.get_by_task("task-2") is None
+
+
+def test_duplicate_task_id_raises_integrity_error(store: SQLiteVariantStore) -> None:
+    # task_id is the grading/telemetry trace key — globally unique: a
+    # second variant may never claim an existing trace key, even under a
+    # different (learner_id, template_id) pair.
+    store.save(make_variant(learner_id="learner-1", template_id="template-1"))
+
+    with pytest.raises(sa.exc.IntegrityError):
+        store.save(
+            make_variant(
+                learner_id="learner-2",
+                template_id="template-2",
+                task_id="task-1",  # collides with learner-1's trace key
+            )
+        )
+
+    # Rejected row not partially stored under either identity.
+    assert store.get("learner-2", "template-2") is None
+    assert len(store.list_by_template("template-2")) == 0
+
+
+def test_list_for_learner_roundtrip_and_audit(store: SQLiteVariantStore) -> None:
+    # Created out of insertion order; list must come back chronological.
+    store.save(make_variant(template_id="template-c", task_id="task-c",
+                            created_at=_BASE_TS + timedelta(hours=2)))
+    store.save(make_variant(template_id="template-a", task_id="task-a",
+                            created_at=_BASE_TS))
+    store.save(make_variant(template_id="template-b", task_id="task-b",
+                            created_at=_BASE_TS + timedelta(hours=1)))
+
+    variants = store.list_for_learner("learner-1")
+    assert [v.template_id for v in variants] == [
+        "template-a",
+        "template-b",
+        "template-c",
+    ]
+    assert all(v.learner_id == "learner-1" for v in variants)
+    hours = (timedelta(hours=0), timedelta(hours=1), timedelta(hours=2))
+    assert all(
+        v.created_at == _BASE_TS + offset
+        for v, offset in zip(variants, hours, strict=True)
+    )
+    assert all(v.created_at.tzinfo is UTC for v in variants)
+
+    # Auditable: every stored variant reads back its seed and typed params
+    # (the proctoring cross-check path reads exactly this).
+    for v in variants:
+        assert v.seed.startswith("seed-")
+        assert v.params["scenario"] == "cache invalidation"
+        assert "streaming" in v.params["constraints"]
+        assert v.params["data_shape"]["rows"] == 10_000
+        assert v.starter_files["tests/test_stream_cache.py"].count("\n") >= 1
+
+
+def test_list_by_template_roundtrip_and_audit(store: SQLiteVariantStore) -> None:
+    # Three learners on the same template: distinct, auditable variants.
+    store.save(make_variant(learner_id="learner-b", task_id="task-b",
+                            seed="seed-222",
+                            created_at=_BASE_TS + timedelta(hours=1)))
+    store.save(make_variant(learner_id="learner-a", task_id="task-a",
+                            seed="seed-111", created_at=_BASE_TS))
+    store.save(make_variant(learner_id="learner-c", task_id="task-c",
+                            seed="seed-333",
+                            created_at=_BASE_TS + timedelta(hours=2)))
+
+    variants = store.list_by_template("template-1")
+    assert [v.learner_id for v in variants] == ["learner-a", "learner-b", "learner-c"]
+    assert all(v.template_id == "template-1" for v in variants)
+
+    # Every learner's variant carries its own seed + params (auditable,
+    # REQ-3-005: variant parameters persisted and auditable).
+    seeds = {v.learner_id: v.seed for v in variants}
+    assert seeds == {
+        "learner-a": "seed-111",
+        "learner-b": "seed-222",
+        "learner-c": "seed-333",
+    }
+    assert all(v.params["scenario"] == "cache invalidation" for v in variants)
+    assert all(v.created_at.tzinfo is UTC for v in variants)
+
+
+def test_list_unknown_returns_empty_lists(store: SQLiteVariantStore) -> None:
+    store.save(make_variant())
+
+    assert store.list_for_learner("nobody") == []
+    assert store.list_by_template("no-template") == []
+
+
+def test_lists_are_scoped(store: SQLiteVariantStore) -> None:
+    store.save(make_variant(learner_id="learner-1", template_id="template-1",
+                            task_id="task-1"))
+    store.save(make_variant(learner_id="learner-2", template_id="template-1",
+                            task_id="task-2"))
+    store.save(make_variant(learner_id="learner-1", template_id="template-2",
+                            task_id="task-3"))
+
+    # learner lists see only that learner's rows.
+    assert [v.template_id for v in store.list_for_learner("learner-1")] == [
+        "template-1",
+        "template-2",
+    ]
+    assert [v.template_id for v in store.list_for_learner("learner-2")] == ["template-1"]
+
+    # template lists see one row per learner, none from other templates.
+    template_rows = store.list_by_template("template-1")
+    assert sorted(v.learner_id for v in template_rows) == ["learner-1", "learner-2"]
+    assert all(v.template_id == "template-1" for v in template_rows)
+
+    # get stays a pair-scoped point lookup: same template, other learner.
+    assert store.get("learner-1", "template-2") is not None
+    assert store.get("learner-2", "template-2") is None
+
+
+def test_empty_params_and_starter_files_roundtrip(store: SQLiteVariantStore) -> None:
+    # Legal shapes: a slotless template carries no params; a variant may
+    # ship without a workspace scaffold.
+    store.save(make_variant(params={}, starter_files={}))
+
+    fetched = store.get("learner-1", "template-1")
+    assert fetched is not None
+    assert fetched.params == {}
+    assert fetched.starter_files == {}
+
+
+def test_rows_are_detached_after_save(store: SQLiteVariantStore, tmp_path: Path) -> None:
+    # The API layer hands VariantRecords across layers; rows must survive
+    # the store that produced them being closed (no open-session ORM magic).
+    store.save(make_variant())
+    fetched = store.get("learner-1", "template-1")
+    by_task = store.get_by_task("task-1")
+    store.close()
+
+    assert fetched is not None
+    assert fetched.statement == fetched.statement  # usable post-close
+    assert fetched.starter_files["src/stream_cache.py"] == "class StreamCache:\n    pass\n"
+    assert by_task is not None
+    assert by_task.seed == "seed-a1b2c3"
+
+    # A fresh store on the same file sees the same row (durability).
+    reopened = SQLiteVariantStore(db_path=tmp_path / "variants.db")
+    try:
+        again = reopened.get("learner-1", "template-1")
+        assert again is not None
+        assert again.starter_files["src/stream_cache.py"].startswith("class StreamCache")
+        assert again.created_at.tzinfo is UTC
+    finally:
+        reopened.close()
+
+
+def test_pragmas_are_applied(store: SQLiteVariantStore) -> None:
+    # Pragmas are per-connection; query through the store's engine so the
+    # connect hook (not a default sqlite3 connection) is what we inspect.
+    with store._engine.connect() as conn:
+        (journal_mode,) = conn.execute(sa.text("PRAGMA journal_mode")).one()
+        (synchronous,) = conn.execute(sa.text("PRAGMA synchronous")).one()
+    assert journal_mode == "wal"
+    # synchronous=NORMAL is 1 in SQLite's pragma numbering.
+    assert synchronous == 1
+
+
+def test_concurrent_writer_and_reader_no_database_is_locked(tmp_path: Path) -> None:
+    """One thread saves while another reads in a tight loop (a-3).
+
+    Without WAL + busy_timeout this pattern reliably produces
+    `OperationalError: database is locked` on SQLite. The assertion is
+    that every reader call completes and every distinct-row write lands.
+    """
+    db_path = tmp_path / "variants.db"
+    n_variants = 60  # distinct (learner, template) pairs, one save each
+    stop_writing = threading.Event()
+
+    writer = SQLiteVariantStore(db_path=db_path)
+    reader = SQLiteVariantStore(db_path=db_path)
+    try:
+
+        def write_variants() -> None:
+            for seq in range(n_variants):
+                writer.save(
+                    make_variant(
+                        learner_id=f"learner-{seq}",
+                        template_id="template-1",
+                        task_id=f"task-{seq}",
+                        seed=f"seed-{seq:03d}",
+                        created_at=_BASE_TS + timedelta(seconds=seq),
+                    )
+                )
+            stop_writing.set()
+
+        def read_variants() -> None:
+            while not stop_writing.is_set():
+                reader.list_by_template("template-1")
+            # Final read after the writer is done.
+            assert len(reader.list_by_template("template-1")) == n_variants
+
+        with concurrent.futures.ThreadPoolExecutor(max_workers=2) as pool:
+            futures = [pool.submit(write_variants), pool.submit(read_variants)]
+            for future in futures:
+                future.result(timeout=30)
+
+        # Every save was a distinct row: nothing lost, none doubled.
+        assert len(writer.list_by_template("template-1")) == n_variants
+    finally:
+        reader.close()
+        writer.close()

apps/ai-service/tests/variants/test_templates.py

@@ -0,0 +1,102 @@
+"""Task template tests (Task 4-1-01, REQ-3-005)."""
+
+from __future__ import annotations
+
+import random
+
+import pytest
+
+from ai_service.variants.templates import (
+    TEMPLATES,
+    ParameterSlot,
+    TaskTemplate,
+    get_template,
+    slots_pattern_ok,
+    template_for_competency,
+    validate_competency_binding,
+)
+
+
+def test_all_templates_bind_to_real_competency_ids() -> None:
+    validate_competency_binding()  # raises on any unknown binding
+    assert len(TEMPLATES) >= 3
+
+
+def test_slot_validation_rejects_bad_values() -> None:
+    slot = ParameterSlot(name="domain", type="enum", values=["a", "b"])
+    assert slot.validate_value("a")
+    assert not slot.validate_value("c")
+    assert not slot.validate_value(3)
+
+    rng = random.Random(42)
+    assert slot.sample(rng) in {"a", "b"}
+
+
+def test_int_range_slot_bounds() -> None:
+    slot = ParameterSlot(name="n", type="int_range", lo=2, hi=4)
+    rng = random.Random(0)
+    for _ in range(20):
+        assert 2 <= slot.sample(rng) <= 4
+    assert slot.validate_value(3)
+    assert not slot.validate_value(5)
+    assert not slot.validate_value("3")
+
+
+def test_seeded_sampling_is_reproducible() -> None:
+    tpl = TEMPLATES["tpl-llm-judge"]
+    first = tpl.sample_params(seed=1234)
+    second = tpl.sample_params(seed=1234)
+    other = tpl.sample_params(seed=1235)
+    assert first == second  # D-029: same seed -> identical params
+    assert first != other  # different seed -> (near-certainly) different draw
+
+
+def test_skeleton_placeholders_match_slots() -> None:
+    for tpl in TEMPLATES.values():
+        assert slots_pattern_ok(tpl.statement_skeleton, tpl.slots), tpl.id
+
+
+def test_render_validates_params_and_fills() -> None:
+    tpl = TEMPLATES["tpl-llm-judge"]
+    params = tpl.sample_params(seed=7)
+    rendered = tpl.render(params)
+    for value in params.values():
+        assert str(value) in rendered
+    with pytest.raises(ValueError, match="invalid value"):
+        tpl.render({**params, "edge_cases": 99})  # out of band
+
+
+def test_rubric_anchors_present_per_template() -> None:
+    for tpl in TEMPLATES.values():
+        anchors = tpl.rubric_anchors
+        assert anchors.expected_edit_count_band[0] <= anchors.expected_edit_count_band[1]
+        assert anchors.expected_min_test_runs >= 1
+        cycles = anchors.expected_error_fix_cycles_band
+        assert cycles[0] <= cycles[1]
+
+
+def test_starter_files_defined_per_template() -> None:
+    for tpl in TEMPLATES.values():
+        assert tpl.starter_files, f"{tpl.id} missing starter scaffolds"
+        assert "README.md" in tpl.starter_files
+        assert tpl.test_command
+
+
+def test_competency_lookup() -> None:
+    tpls = template_for_competency("stack-orchestration-c005")
+    assert len(tpls) == 1
+    assert get_template("nope-xyz") is None
+
+
+def test_bad_skeleton_rejected() -> None:
+    with pytest.raises(ValueError, match="slot"):
+        TaskTemplate(
+            id="tpl-bad",
+            competency_id="stack-orchestration-c001",
+            title="Bad",
+            statement_skeleton="no placeholders at all",
+            slots=[ParameterSlot(name="x", type="enum", values=["a"])],
+            rubric_anchors=TEMPLATES["tpl-llm-judge"].rubric_anchors,
+            starter_files={},
+            test_command="pytest -q",
+        )

apps/ai-service/tests/voice/__init__.py

@@ -0,0 +1 @@
+"""Voice layer tests — provider mock/fallback + DefenseStore (REQ-3-006)."""

apps/ai-service/tests/voice/test_defense_store.py

@@ -0,0 +1,469 @@
+"""SQLiteDefenseStore tests (REQ-3-006, D-027).
+
+Each test gets its own tmp-path SQLite file — no shared disk state. Covers:
+    - start → append_turn (examiner + learner interleaved) → finalize →
+      get roundtrip: every field survives (including nested JSON
+      integrity signals, per-turn latency_ms, and the tz-aware
+      created_at/ts contract) and turns come back ordered by seq
+    - lifecycle ownership: start forces in_progress + finished_at=None
+      even if the caller smuggles a finished status
+    - insert-only start: a duplicate id raises IntegrityError (the id is
+      minted once per session); the original row is untouched
+    - append_turn validation: duplicate (defense_id, seq) raises
+      (a transcript turn must never silently vanish); unknown
+      defense_id raises (FK enforced); mismatched defense_id argument
+      raises ValueError before touching the DB; negative seq rejected
+    - finalize: unknown id → None (documented behavior — the API maps
+      it to 404); re-finalize is latest-wins on signals + finished_at
+    - get: unknown id → None; turns attached only by get() —
+      list_for_learner records carry turns == []
+    - list_for_learner: scoped per learner, chronological
+    - rows detached: usable after the store is closed; durability across
+      a fresh store on the same file
+    - WAL + synchronous=NORMAL + foreign_keys=ON pragmas actually applied
+"""
+
+from datetime import UTC, datetime, timedelta
+from pathlib import Path
+
+import pytest
+import sqlalchemy as sa
+
+from ai_service.voice.defense_store import (
+    DefenseRecord,
+    DefenseTurn,
+    SQLiteDefenseStore,
+)
+
+_BASE_TS = datetime(2026, 9, 12, 12, 0, 0, tzinfo=UTC)
+
+
+def make_defense(
+    id: str = "defense-1",  # shadows builtin deliberately: DefenseRecord field name
+    learner_id: str = "learner-1",
+    task_id: str = "task-1",
+    status: str = "in_progress",
+    created_at: datetime | None = None,
+) -> DefenseRecord:
+    """Canonical kwargs builder — tests override only what they assert on."""
+    return DefenseRecord(
+        id=id,
+        learner_id=learner_id,
+        task_id=task_id,
+        status=status,
+        created_at=created_at if created_at is not None else _BASE_TS,
+    )
+
+
+def make_turn(
+    defense_id: str = "defense-1",
+    seq: int = 0,
+    role: str = "examiner",
+    text: str = "Walk me through your cache invalidation strategy.",
+    ts: datetime | None = None,
+    latency_ms: int | None = 240,
+    created_at: datetime | None = None,
+) -> DefenseTurn:
+    return DefenseTurn(
+        defense_id=defense_id,
+        seq=seq,
+        role=role,
+        text=text,
+        ts=ts if ts is not None else _BASE_TS + timedelta(seconds=seq),
+        latency_ms=latency_ms,
+        created_at=created_at
+        if created_at is not None
+        else _BASE_TS + timedelta(seconds=seq),
+    )
+
+
+@pytest.fixture
+def store(tmp_path: Path) -> SQLiteDefenseStore:
+    s = SQLiteDefenseStore(db_path=tmp_path / "defenses.db")
+    yield s
+    s.close()
+
+
+def test_start_append_finalize_get_roundtrip(store: SQLiteDefenseStore) -> None:
+    """THE roundtrip of the defense lifecycle (REQ-3-006)."""
+    # start: a defense is born in_progress with empty signals.
+    started = store.start(make_defense())
+    assert started.status == "in_progress"
+    assert started.finished_at is None
+    assert started.integrity_signals == {}
+    assert started.created_at == _BASE_TS
+    assert started.created_at.tzinfo is UTC
+
+    # append: examiner + learner turns interleaved — append them OUT of
+    # seq order to prove get() orders by seq, not by insertion.
+    learner_a = make_turn(
+        seq=1, role="learner", text="I invalidate on write-ahead flush.", latency_ms=980
+    )
+    examiner_b = make_turn(
+        seq=2, role="examiner", text="Why not invalidate on read?", latency_ms=180
+    )
+    learner_c = make_turn(
+        seq=3, role="learner", text="Read-path misses were rare in my trace.", latency_ms=1100
+    )
+    store.append_turn("defense-1", make_turn(seq=0))  # first examiner question
+    store.append_turn("defense-1", learner_a)
+    store.append_turn("defense-1", examiner_b)
+    store.append_turn("defense-1", learner_c)
+
+    # finalize: seal with A-109 integrity signals (long pauses, off-scope).
+    finalize_started = datetime.now(UTC)  # real wall clock, not the fixture
+    signals = {
+        "long_pauses": {"count": 2, "threshold_ms": 2000, "turns": [1, 3]},
+        "off_scope": {"count": 1, "turns": [3], "markers": ["unrelated tangent"]},
+        "verdict": "PASS_WITH_NOTES",
+    }
+    finalized = store.finalize("defense-1", signals)
+    assert finalized is not None
+    assert finalized.status == "finished"
+    assert finalized.finished_at is not None
+    assert finalized.finished_at.tzinfo is UTC
+    # finalize() does not attach turns — get() is the with-turns path.
+    assert finalized.turns == []
+
+    # get: full transcript in seq order + signals persisted.
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.id == "defense-1"
+    assert fetched.learner_id == "learner-1"
+    assert fetched.task_id == "task-1"
+    assert fetched.status == "finished"
+    assert fetched.integrity_signals == signals
+    assert fetched.integrity_signals["long_pauses"]["turns"] == [1, 3]  # nested JSON survives
+    assert fetched.created_at == _BASE_TS
+    assert fetched.created_at.tzinfo is UTC
+    assert fetched.finished_at is not None
+    assert fetched.finished_at >= finalize_started  # stamped at finalize time
+
+    # THE ordered-transcript assertion.
+    assert [t.seq for t in fetched.turns] == [0, 1, 2, 3]
+    assert [t.role for t in fetched.turns] == [
+        "examiner",
+        "learner",
+        "examiner",
+        "learner",
+    ]
+    assert fetched.turns[0].text == "Walk me through your cache invalidation strategy."
+    assert fetched.turns[1].latency_ms == 980
+    assert fetched.turns[2].latency_ms == 180
+    assert fetched.turns[3].latency_ms == 1100
+    # Per-turn ts contract: tz-aware UTC on read regardless of backend.
+    assert all(t.ts.tzinfo is UTC for t in fetched.turns)
+    assert all(t.created_at.tzinfo is UTC for t in fetched.turns)
+    assert [t.defense_id for t in fetched.turns] == ["defense-1"] * 4
+
+
+def test_start_forces_in_progress_lifecycle(store: SQLiteDefenseStore) -> None:
+    """The store owns the lifecycle: a smuggled finished status is
+    normalized away at birth — only finalize() may move a defense to
+    finished."""
+    smuggled = make_defense(status="finished")
+    smuggled.finished_at = _BASE_TS + timedelta(hours=1)
+    started = store.start(smuggled)
+
+    assert started.status == "in_progress"
+    assert started.finished_at is None
+
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.status == "in_progress"
+    assert fetched.finished_at is None
+
+
+def test_duplicate_defense_id_raises_integrity_error(
+    store: SQLiteDefenseStore,
+) -> None:
+    """Insert-only start (first-wins): the id is minted once per session;
+    a duplicate is a bug or lost race to surface, not swallow."""
+    store.start(make_defense(learner_id="learner-1"))
+    store.start(make_defense(id="defense-2", learner_id="learner-2"))
+
+    with pytest.raises(sa.exc.IntegrityError):
+        store.start(make_defense(learner_id="learner-3"))  # same id again
+
+    # The original rows survived intact — nothing was overwritten.
+    first = store.get("defense-1")
+    assert first is not None
+    assert first.learner_id == "learner-1"
+    assert first.status == "in_progress"
+    second = store.get("defense-2")
+    assert second is not None
+    assert second.learner_id == "learner-2"
+
+
+def test_append_turn_duplicate_seq_raises_integrity_error(
+    store: SQLiteDefenseStore,
+) -> None:
+    """A transcript turn must never silently vanish: (defense_id, seq) is
+    the PK, so a duplicate raises instead of overwriting."""
+    store.start(make_defense())
+    store.append_turn("defense-1", make_turn(seq=0))
+    store.append_turn("defense-1", make_turn(seq=1))
+
+    with pytest.raises(sa.exc.IntegrityError):
+        store.append_turn(
+            "defense-1",
+            make_turn(seq=1, role="learner", text="an impostor answer"),
+        )
+
+    # The stored turn is untouched — NOT upsert.
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert len(fetched.turns) == 2
+    assert fetched.turns[1].role == "examiner"
+    assert fetched.turns[1].text != "an impostor answer"
+
+
+def test_append_turn_unknown_defense_raises_integrity_error(
+    store: SQLiteDefenseStore,
+) -> None:
+    """FK enforced (foreign_keys=ON): an orphan turn is rejected, not
+    silently attached to a defense that does not exist."""
+    store.start(make_defense())
+
+    with pytest.raises(sa.exc.IntegrityError):
+        store.append_turn(
+            "defense-missing",
+            make_turn(defense_id="defense-missing", seq=0),
+        )
+
+    # And the turn did not land on the existing defense either.
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.turns == []
+
+
+def test_append_turn_identity_mismatch_raises_value_error(
+    store: SQLiteDefenseStore,
+) -> None:
+    """The defense_id argument is the write identity: a turn object
+    claiming another defense is a programming error — surfaced BEFORE
+    any DB round-trip."""
+    store.start(make_defense())
+
+    with pytest.raises(ValueError, match="does not match"):
+        store.append_turn(
+            "defense-1",
+            make_turn(defense_id="defense-other", seq=0),
+        )
+
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.turns == []
+
+
+def test_turn_model_rejects_negative_seq_and_unknown_role(store: SQLiteDefenseStore) -> None:
+    """Model-level contracts (@validates hooks fire at construction)."""
+    with pytest.raises(ValueError, match="seq"):
+        make_turn(seq=-1)
+    with pytest.raises(ValueError, match="role"):
+        make_turn(role="proctor")
+    with pytest.raises(ValueError, match="text"):
+        make_turn(text="")
+    with pytest.raises(ValueError, match="latency_ms"):
+        make_turn(latency_ms=-5)
+    with pytest.raises(ValueError, match="defense_id"):
+        make_turn(defense_id="")
+
+
+def test_append_turn_allows_missing_latency(store: SQLiteDefenseStore) -> None:
+    """latency_ms is None until the endpoints instrument it (task
+    5-4-01) — None must roundtrip cleanly."""
+    store.start(make_defense())
+    store.append_turn(
+        "defense-1",
+        make_turn(seq=0, role="examiner", latency_ms=None),
+    )
+
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.turns[0].latency_ms is None
+
+
+def test_finalize_unknown_id_returns_none(store: SQLiteDefenseStore) -> None:
+    """Documented unknown-id behavior: None, not a raise — the defense
+    endpoints (task 5-3-01) map this straight to 404."""
+    assert store.finalize("nobody", {"verdict": "PASS"}) is None
+
+
+def test_finalize_is_latest_wins_on_refinalize(store: SQLiteDefenseStore) -> None:
+    """A recomputed verdict replaces the stored one wholesale (mirrors
+    GradeStore.save): signals + finished_at are overwritten, status
+    just stays finished."""
+    store.start(make_defense())
+    first_signals = {"verdict": "FAIL", "long_pauses": {"count": 5}}
+    store.finalize("defense-1", first_signals)
+
+    better_signals = {
+        "verdict": "PASS",
+        "long_pauses": {"count": 1},
+        "off_scope": {"count": 0},
+    }
+    refinalized = store.finalize("defense-1", better_signals)
+    assert refinalized is not None
+
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.status == "finished"
+    assert fetched.integrity_signals == better_signals  # wholesale replace
+    assert fetched.integrity_signals["verdict"] == "PASS"
+    assert refinalized.finished_at is not None
+    assert fetched.finished_at == refinalized.finished_at  # stamped anew
+
+
+def test_finalize_rejects_non_dict_signals(store: SQLiteDefenseStore) -> None:
+    """The signals column contract is a JSON OBJECT dict; None/str/list
+    would break every reader (Examiner feed, Proctor/Mentor)."""
+    store.start(make_defense())
+
+    with pytest.raises(ValueError, match="integrity_signals"):
+        store.finalize("defense-1", None)  # type: ignore[arg-type]
+
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.status == "in_progress"  # untouched by the rejected call
+
+
+def test_get_unknown_returns_none(store: SQLiteDefenseStore) -> None:
+    store.start(make_defense())
+
+    assert store.get("defense-1") is not None
+    assert store.get("defense-missing") is None
+
+
+def test_empty_signals_until_finalize(store: SQLiteDefenseStore) -> None:
+    """A-109: signals are {} until finalize — the in-progress transcript
+    is readable without any verdict present."""
+    store.start(make_defense())
+    store.append_turn("defense-1", make_turn(seq=0))
+
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    assert fetched.status == "in_progress"
+    assert fetched.integrity_signals == {}
+    assert fetched.finished_at is None
+    assert len(fetched.turns) == 1
+
+
+def test_list_for_learner_scoped_and_chronological(store: SQLiteDefenseStore) -> None:
+    # Created out of insertion order; list must come back chronological.
+    store.start(make_defense(id="defense-c", learner_id="learner-1",
+                             created_at=_BASE_TS + timedelta(hours=2)))
+    store.start(make_defense(id="defense-a", learner_id="learner-1"))
+    store.start(make_defense(id="defense-b", learner_id="learner-2"))
+    # Turns on learner-1's defenses prove the list carries NONE of them.
+    store.append_turn("defense-c", make_turn(defense_id="defense-c", seq=0))
+    store.append_turn("defense-a", make_turn(defense_id="defense-a", seq=0))
+
+    defenses = store.list_for_learner("learner-1")
+    assert [d.id for d in defenses] == ["defense-a", "defense-c"]
+    assert all(d.learner_id == "learner-1" for d in defenses)
+    # WITHOUT turns: the list feed carries session headers only — get()
+    # is the with-turns path.
+    assert all(d.turns == [] for d in defenses)
+    # Headers intact: status + signals readable for the Proctor/Mentor feed.
+    assert all(d.status == "in_progress" for d in defenses)
+    assert all(d.integrity_signals == {} for d in defenses)
+    assert all(d.created_at.tzinfo is UTC for d in defenses)
+
+    # Other learners never leak.
+    assert [d.id for d in store.list_for_learner("learner-2")] == ["defense-b"]
+    assert store.list_for_learner("learner-missing") == []
+
+
+def test_list_includes_finalized_with_signals(store: SQLiteDefenseStore) -> None:
+    """The learner feed must surface finished defenses WITH their sealed
+    signals (the Proctor/Mentor cross-check reads exactly this)."""
+    store.start(make_defense())
+    signals = {"verdict": "PASS", "long_pauses": {"count": 0}}
+    store.finalize("defense-1", signals)
+
+    defenses = store.list_for_learner("learner-1")
+    assert len(defenses) == 1
+    assert defenses[0].status == "finished"
+    assert defenses[0].integrity_signals == signals
+    assert defenses[0].finished_at is not None
+    assert defenses[0].finished_at.tzinfo is UTC
+
+
+def test_rows_are_detached_and_durable(store: SQLiteDefenseStore, tmp_path: Path) -> None:
+    """Detached from any session: the API layer hands DefenseRecords
+    across layers; rows must survive the store that produced them
+    being closed, and a fresh store must see the same rows."""
+    store.start(make_defense())
+    store.append_turn("defense-1", make_turn(seq=0))
+    store.append_turn("defense-1", make_turn(seq=1, role="learner", text="My answer."))
+    store.finalize("defense-1", {"verdict": "PASS"})
+    fetched = store.get("defense-1")
+    assert fetched is not None
+    store.close()
+
+    # Usable post-close — no open-session ORM magic.
+    assert fetched.status == "finished"
+    assert fetched.integrity_signals["verdict"] == "PASS"
+    assert [t.text for t in fetched.turns] == [
+        "Walk me through your cache invalidation strategy.",
+        "My answer.",
+    ]
+
+    # A fresh store on the same file sees the same rows (durability).
+    reopened = SQLiteDefenseStore(db_path=tmp_path / "defenses.db")
+    try:
+        again = reopened.get("defense-1")
+        assert again is not None
+        assert again.status == "finished"
+        assert again.integrity_signals == {"verdict": "PASS"}
+        assert [t.seq for t in again.turns] == [0, 1]
+        assert again.turns[1].latency_ms == 240
+        assert all(t.ts.tzinfo is UTC for t in again.turns)
+    finally:
+        reopened.close()
+
+
+def test_pragmas_are_applied(store: SQLiteDefenseStore) -> None:
+    # Pragmas are per-connection; query through the store's engine so the
+    # connect hook (not a default sqlite3 connection) is what we inspect.
+    with store._engine.connect() as conn:
+        (journal_mode,) = conn.execute(sa.text("PRAGMA journal_mode")).one()
+        (synchronous,) = conn.execute(sa.text("PRAGMA synchronous")).one()
+        (foreign_keys,) = conn.execute(sa.text("PRAGMA foreign_keys")).one()
+    assert journal_mode == "wal"
+    # synchronous=NORMAL is 1 in SQLite's pragma numbering.
+    assert synchronous == 1
+    # The FK is actually enforced on SQLite (Postgres parity, D-027).
+    assert foreign_keys == 1
+
+
+def test_two_defenses_same_learner_independent_transcripts(
+    store: SQLiteDefenseStore,
+) -> None:
+    """Scoped transcripts: two defenses never see each other's turns."""
+    store.start(make_defense(id="defense-a", task_id="task-1"))
+    store.start(make_defense(id="defense-b", task_id="task-2"))
+    # Both defenses reuse seq 0,1 — per-defense numbering.
+    for defense_id in ("defense-a", "defense-b"):
+        store.append_turn(defense_id, make_turn(defense_id=defense_id, seq=0))
+        store.append_turn(
+            defense_id,
+            make_turn(
+                defense_id=defense_id, seq=1, role="learner",
+                text=f"answer for {defense_id}",
+            ),
+        )
+
+    a = store.get("defense-a")
+    b = store.get("defense-b")
+    assert a is not None and b is not None
+    assert [t.seq for t in a.turns] == [0, 1]
+    assert [t.text for t in a.turns] == [
+        "Walk me through your cache invalidation strategy.",
+        "answer for defense-a",
+    ]
+    assert [t.text for t in b.turns] == [
+        "Walk me through your cache invalidation strategy.",
+        "answer for defense-b",
+    ]

apps/ai-service/tests/voice/test_latency.py

@@ -0,0 +1,101 @@
+"""Per-turn latency instrumentation tests (Task 5-4-01, REQ-3-006, A-109).
+
+Mock-based: asserts instrumentation PRESENCE and population (stt_ms / llm_ms /
+tts_ms fields, per-turn latency_ms persisted, the budget constant defined) —
+wall-clock against a real voice endpoint is a v0.4 acceptance criterion
+(real STT/TTS deferred per GRILL CUT-1 / G-7).
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from ai_service.agents.examiner import ExaminerAgent
+from ai_service.config import Settings
+from ai_service.grading.store import SQLiteGradeStore
+from ai_service.llm.mock import MockProvider
+from ai_service.main import create_app
+from ai_service.telemetry.ingest import TraceIntegrityMap
+from ai_service.telemetry.store import SQLiteTraceStore
+from ai_service.variants.store import SQLiteVariantStore
+from ai_service.voice.defense_store import SQLiteDefenseStore
+from ai_service.voice.mock import MockVoiceProvider
+
+#: A-109: the documented conversational budget (acceptance criterion for the
+#: v0.4 real-voice probe; mock turns are near-instant so v0.3 asserts
+#: instrumentation, not wall-clock).
+DEFENSE_TURN_BUDGET_MS = 4_000
+
+
+@pytest.fixture()
+def client(tmp_path: Path) -> TestClient:
+    llm = MockProvider()
+    app = create_app(Settings(provider="mock", voice_provider="mock"))
+    app.state.provider = llm
+    app.state.trace_store = SQLiteTraceStore(db_path=tmp_path / "t.db")
+    app.state.variant_store = SQLiteVariantStore(db_path=tmp_path / "v.db")
+    app.state.grade_store = SQLiteGradeStore(db_path=tmp_path / "g.db")
+    app.state.trace_integrity = TraceIntegrityMap()
+    app.state.defense_store = SQLiteDefenseStore(db_path=tmp_path / "d.db")
+    app.state.voice_provider = MockVoiceProvider(["my answer"])
+    app.state.examiner_agent = ExaminerAgent(llm, Settings(provider="mock"))
+    with TestClient(app) as c:
+        yield c
+
+
+class TestLatencyInstrumentation:
+    def test_budget_constant_defined(self) -> None:
+        """The conversational budget is a named, documented constant (A-109)."""
+        assert DEFENSE_TURN_BUDGET_MS > 0
+        assert DEFENSE_TURN_BUDGET_MS <= 5_000  # conversational feel target
+
+    def test_answer_reports_per_phase_latency(self, client: TestClient) -> None:
+        start = client.post(
+            "/v1/defense/start",
+            json={"learner_id": "lat-learner", "task_id": "lat-task"},
+        ).json()
+        resp = client.post(
+            f"/v1/defense/{start['defense_id']}/answer", data={"text": "answer"}
+        )
+        assert resp.status_code == 200
+        latency = resp.json()["turn_latency"]
+        assert latency["llm_ms"] is not None and latency["llm_ms"] >= 0
+        assert "stt_ms" in latency and "tts_ms" in latency
+
+    def test_audio_answer_populates_stt_ms(self, client: TestClient) -> None:
+        start = client.post(
+            "/v1/defense/start",
+            json={"learner_id": "lat-learner", "task_id": "lat-task"},
+        ).json()
+        resp = client.post(
+            f"/v1/defense/{start['defense_id']}/answer",
+            files={"audio": ("a.wav", b"RIFF" + b"\x00" * 32, "audio/wav")},
+        )
+        latency = resp.json()["turn_latency"]
+        assert latency["stt_ms"] is not None and latency["stt_ms"] >= 0
+
+    def test_every_turn_persists_latency_ms(self, client: TestClient) -> None:
+        start = client.post(
+            "/v1/defense/start",
+            json={"learner_id": "lat-learner", "task_id": "lat-task"},
+        ).json()
+        client.post(f"/v1/defense/{start['defense_id']}/answer", data={"text": "a"})
+        transcript = client.get(f"/v1/defense/{start['defense_id']}").json()
+        assert transcript["turns"]
+        for turn in transcript["turns"]:
+            assert "latency_ms" in turn
+            assert turn["latency_ms"] is not None or turn["role"] == "learner"
+
+    def test_mock_turns_within_budget(self, client: TestClient) -> None:
+        """Mock turns must be near-instant — the budget holds trivially."""
+        start = client.post(
+            "/v1/defense/start",
+            json={"learner_id": "lat-learner", "task_id": "lat-task"},
+        ).json()
+        resp = client.post(
+            f"/v1/defense/{start['defense_id']}/answer", data={"text": "a"}
+        ).json()
+        assert resp["turn_latency"]["llm_ms"] < DEFENSE_TURN_BUDGET_MS

apps/ai-service/tests/voice/test_voice_layer.py

@@ -0,0 +1,118 @@
+"""Voice layer tests (Task 5-1-01, REQ-3-006) — D-030 mock-first, zero network."""
+
+from __future__ import annotations
+
+import pytest
+
+from ai_service.config import Settings
+from ai_service.voice.browser import BROWSER_FALLBACK_DESCRIPTOR, MOCK_DESCRIPTOR
+from ai_service.voice.factory import UnknownVoiceProviderError, voice_provider_from_settings
+from ai_service.voice.mock import MockVoiceFailure, MockVoiceProvider, _tone_wav
+
+
+class TestMockVoiceProvider:
+    async def test_transcribe_deterministic(self) -> None:
+        provider = MockVoiceProvider(["hello defense"])
+        a = await provider.transcribe(b"x" * 3200, "wav")
+        b = await provider.transcribe(b"x" * 3200, "wav")
+        assert a.text == b.text == "hello defense"
+        assert a.duration_ms == b.duration_ms
+
+    async def test_transcribe_canned_queue(self) -> None:
+        provider = MockVoiceProvider(["first answer", "second answer"])
+        first = await provider.transcribe(b"audio", "webm")
+        second = await provider.transcribe(b"audio", "webm")
+        assert first.text == "first answer"
+        assert second.text == "second answer"
+
+    async def test_transcribe_empty_audio_fails(self) -> None:
+        provider = MockVoiceProvider(["x"])
+        with pytest.raises(MockVoiceFailure, match="no audio"):
+            await provider.transcribe(b"", "wav")
+
+    async def test_transcribe_scripted_failure_mode(self) -> None:
+        provider = MockVoiceProvider(["FAIL"])
+        with pytest.raises(MockVoiceFailure, match="scripted STT failure"):
+            await provider.transcribe(b"audio", "wav")
+
+    async def test_synthesize_yields_nonempty_chunks(self) -> None:
+        provider = MockVoiceProvider()
+        chunks = [chunk async for chunk in provider.synthesize("question text")]
+        assert chunks
+        assert all(isinstance(c, bytes) and c for c in chunks)
+        assert provider.synthesize_calls == 1
+
+    async def test_synthesize_empty_text_fails(self) -> None:
+        provider = MockVoiceProvider()
+        with pytest.raises(MockVoiceFailure):
+            async for _ in provider.synthesize(""):
+                pass
+
+    def test_tone_wav_is_real_wav(self) -> None:
+        import io
+        import wave
+
+        raw = _tone_wav(duration_ms=100)
+        with wave.open(io.BytesIO(raw)) as w:
+            assert w.getnchannels() == 1
+            assert w.getsampwidth() == 2
+            assert w.getframerate() == 8000
+
+    async def test_identical_synthesize_calls_identical_bytes(self) -> None:
+        p1, p2 = MockVoiceProvider(), MockVoiceProvider()
+        c1 = b"".join([c async for c in p1.synthesize("same text")])
+        c2 = b"".join([c async for c in p2.synthesize("same text")])
+        assert c1 == c2
+
+
+class TestFactory:
+    def test_default_is_mock(self) -> None:
+        provider = voice_provider_from_settings(Settings())
+        assert isinstance(provider, MockVoiceProvider)
+
+    def test_explicit_mock(self) -> None:
+        provider = voice_provider_from_settings(Settings(voice_provider="mock"))
+        assert isinstance(provider, MockVoiceProvider)
+
+    def test_browser_mode_selects_server_side_mock_for_text_fallback(self) -> None:
+        # Browser mode composes the same deterministic provider server-side;
+        # the descriptor tells the CLIENT to use native SR/TTS.
+        provider = voice_provider_from_settings(Settings(voice_provider="browser"))
+        assert isinstance(provider, MockVoiceProvider)
+
+    def test_real_server_stt_tts_rejected_as_v04_seam(self) -> None:
+        with pytest.raises(UnknownVoiceProviderError, match="v0.4"):
+            voice_provider_from_settings(Settings(voice_provider="openai-audio"))
+
+    def test_unknown_provider_rejected(self) -> None:
+        with pytest.raises(UnknownVoiceProviderError, match="unknown"):
+            voice_provider_from_settings(Settings(voice_provider="watson"))
+
+
+class TestDescriptors:
+    def test_browser_fallback_descriptor(self) -> None:
+        assert BROWSER_FALLBACK_DESCRIPTOR.mode == "browser"
+        assert BROWSER_FALLBACK_DESCRIPTOR.sr_available
+        assert BROWSER_FALLBACK_DESCRIPTOR.tts_available
+        assert "SpeechRecognition" in BROWSER_FALLBACK_DESCRIPTOR.hint
+
+    def test_mock_descriptor(self) -> None:
+        assert MOCK_DESCRIPTOR.mode == "mock"
+        assert "v0.4" in MOCK_DESCRIPTOR.hint
+
+
+class TestZeroNetwork:
+    def test_voice_package_never_imports_agents_or_api(self) -> None:
+        import ast
+        from pathlib import Path
+
+        pkg = Path(__file__).parents[2] / "ai_service" / "voice"
+        for py in pkg.glob("*.py"):
+            tree = ast.parse(py.read_text())
+            for node in ast.walk(tree):
+                if isinstance(node, ast.ImportFrom) and node.module:
+                    assert not node.module.startswith("ai_service.agents"), py
+                    assert not node.module.startswith("ai_service.api"), py
+                    if node.level and node.module:
+                        assert node.module.split(".")[-1] != "agents", py
+                        assert node.module.split(".")[-1] != "api", py

apps/cli/README.md

@@ -0,0 +1,48 @@
+# @nextcraft/cli — nextcraft
+
+The bootstrap CLI for the Nextcraft monorepo, shipped as a self-contained linux x64 binary (Node SEA) on every release.
+
+## Commands
+
+See the [root README quickstart](../../README.md) for the user-facing flow. Internals:
+
+- `src/index.ts` — argv dispatch, exit-code contract (0 ok / 1 failure / 2 usage), direct-run guard (`argv[0] === argv[1]` detects SEA context — the installer renames the binary, so filename matching is unreliable)
+- `src/commands/` — doctor / bootstrap / verify / dev; all orchestration delegates to `apps/ai-service/scripts/*.sh` via `src/lib/spawn.ts` (array-args only, SIGTERM→SIGKILL timeout ladder)
+- `src/checks/` — pure logic: version compare, `.env` template diff
+- `tests/` — node:test suites: dispatch, checks, spawn, command stubs, real-box doctor integration, install.sh fixture-server E2E (tamper rejection, degradation), release-assets token isolation, fresh-clone E2E
+
+## Build
+
+```sh
+pnpm cli:typecheck          # tsc --noEmit
+pnpm cli:test               # node:test suites
+pnpm cli:build              # tsc -p tsconfig.build.json -> dist/
+pnpm --filter @nextcraft/cli build:binary <tag>   # SEA binary + sha256 sidecar
+```
+
+`build:binary <tag>`: esbuild bundle (CJS, node18 target, version stamped via `NEXTCRAFT_VERSION_STAMP` define — `--version` reports the tag it was built as) → `node --experimental-sea-config` → postject injection into a copy of the system node binary → `dist/nextcraft-linux-x64` + `dist/nextcraft-linux-x64.sha256`. The binary runs without node on PATH (runtime embedded, ~117 MB).
+
+## Release pipeline
+
+Every ship from v0.3.2 onward runs `scripts/release-assets.sh <tag>` after tag+merge:
+
+1. Builds the binary stamped with the tag
+2. Resolves `GITEA_TOKEN` from `.env*` files ONLY (`.ciagent/.env.secrets` first) — never from shell env
+3. Attaches `nextcraft-linux-x64` + `nextcraft-linux-x64.sha256` to the Gitea release (bounded retry, best-effort — never blocks the ship)
+
+`scripts/install.sh` (POSIX sh, dash-safe): platform gate → Gitea latest-release API resolve → exact-name asset match → sha256 verify BEFORE install (mismatch = hard stop) → `~/.local/bin` install → PATH hint. Any failure degrades to printed source-bootstrap instructions.
+
+## Secrets policy
+
+The CLI never generates, writes, or echoes secrets. `bootstrap` copies `.env.example` → `.env` only when absent and warns on missing optional keys (mock providers keep the stack runnable keyless). Real keys live only in gitignored `.ciagent/.env.secrets`, exported by `apps/ai-service/scripts/dev.sh`.
+
+## Troubleshooting
+
+| Symptom | Cause / fix |
+|---------|-------------|
+| `pnpm not found` in doctor | `corepack enable pnpm` (installs to ~/.local/bin — ensure PATH includes it) |
+| doctor passes but verify fails on venv | re-run `nextcraft bootstrap` (venv/pip resolution is idempotent) |
+| `port 8420 busy` in verify | stop the process on :8420 (`kill $(lsof -t -i:8420)`) or set `AI_PORT` |
+| install.sh says "no binary assets yet" | release predates the binary pipeline (pre-v0.3.2); use source bootstrap |
+| Binary silent after rename | fixed since v0.3.2 (SEA argv detection); re-download the latest release |
+| Checksum mismatch on install | do NOT run the download; delete it and retry — report if it persists |

apps/cli/package.json

@@ -0,0 +1,22 @@
+{
+  "name": "@nextcraft/cli",
+  "version": "0.0.0",
+  "private": true,
+  "type": "module",
+  "bin": {
+    "nextcraft": "dist/index.js"
+  },
+  "scripts": {
+    "dev": "tsx src/index.ts",
+    "test": "tsx --test tests/*.test.ts",
+    "typecheck": "tsc --noEmit",
+    "build": "tsc -p tsconfig.build.json",
+    "build:binary": "node scripts/build-binary.mjs"
+  },
+  "devDependencies": {
+    "@types/node": "^24.0.0",
+    "tsx": "^4.23.0",
+    "typescript": "^5.7.2",
+    "esbuild": "0.28.2"
+  }
+}

apps/cli/scripts/build-binary.mjs

@@ -0,0 +1,63 @@
+#!/usr/bin/env node
+import { execFileSync } from "node:child_process";
+import { copyFileSync, existsSync, readFileSync, rmSync, statSync, writeFileSync } from "node:fs";
+import { createHash } from "node:crypto";
+import { dirname, join } from "node:path";
+import { fileURLToPath } from "node:url";
+
+const pkgDir = dirname(fileURLToPath(import.meta.url)) + "/..";
+const dist = join(pkgDir, "dist");
+const bundle = join(dist, "bundle.cjs");
+const blob = join(dist, "sea-prep.blob");
+const config = join(dist, "sea-config.json");
+const out = join(dist, "nextcraft-linux-x64");
+const checksum = out + ".sha256";
+const version = process.argv[2] ?? "0.0.0-dev";
+
+if (version !== "0.0.0-dev" && !/^v?\d/.test(version)) {
+  console.error(`refusing to stamp implausible version: ${version}`);
+  process.exit(1);
+}
+
+rmSync(dist, { recursive: true, force: true });
+execFileSync(
+  join(pkgDir, "node_modules/.bin/esbuild"),
+  [
+    join(pkgDir, "src/index.ts"),
+    "--bundle",
+    "--platform=node",
+    "--format=cjs",
+    "--target=node18",
+    `--define:NEXTCRAFT_VERSION_STAMP=${JSON.stringify(version)}`,
+    "--outfile=" + bundle,
+  ],
+  { stdio: "inherit" },
+);
+
+const seaConfig = {
+  main: bundle,
+  output: blob,
+  disableExperimentalSEAWarning: true,
+};
+writeFileSync(config, JSON.stringify(seaConfig));
+execFileSync(process.execPath, ["--experimental-sea-config", config], { stdio: "inherit" });
+
+const nodeBin = process.execPath;
+copyFileSync(nodeBin, out);
+execFileSync(
+  "npx",
+  ["--yes", "postject", out, "NODE_SEA_BLOB", blob, "--sentinel-fuse", "NODE_SEA_FUSE_fce680ab2cc467b6e072b8b5df1996b2"],
+  { stdio: "inherit" },
+);
+execFileSync("chmod", ["+x", out]);
+
+const size = statSync(out).size;
+const hash = createHash("sha256").update(readFileSync(out)).digest("hex");
+writeFileSync(checksum, `${hash}  nextcraft-linux-x64\n`);
+
+console.log(`built ${out} (${(size / 1024 / 1024).toFixed(1)} MB) stamped ${version}`);
+console.log(`checksum ${checksum}: ${hash}`);
+if (!existsSync(out) || !existsSync(checksum)) {
+  console.error("expected artifacts missing");
+  process.exit(1);
+}