# COV-001 evidence pack — canonical content-slot schema

## Work-package charter

- **Task ID:** `COV-001`
- **Owner:** Oshun portfolio content-coverage contract maintainer;
  implementation owner: Codex
- **Accountable product:** Shared V1–V9 autonomous-content platform
- **Priority:** P0
- **Target milestone:** 2026-07-15 SOTA gap-closure program
- **Risk tier:** High release-policy, ownership, and coverage-truth integrity
- **Dependencies:** AUD-002 evidence freshness, AUD-003 reproducible
  verification, AUD-006 generated capability truth, AUD-008 reconciled canonical
  audit, Zod, and the shared `@oshun/contracts` package
- **Affected schemas and content slots:** New strict
  `CanonicalContentSlotV1Schema`, explicit legacy v0 input/migration policy,
  stable identity and ownership contracts, and package/root export surfaces. No
  registry rows or shipped content slots are created by COV-001; COV-002 owns
  mechanical seeding.
- **Rollout strategy:** Publish through `@oshun/contracts/content-coverage` and
  the root contracts namespace, require the focused contract/typecheck gate in
  CI, migrate legacy rows only with explicit owner, risk, accessibility, and
  evidence-freshness policy, then seed the registry under COV-002.
- **Rollback strategy:** Revert the schema, exports, package script, CI
  workflow, capability commentary, and audit qualification together. Do not
  import v0 rows through permissive defaults or reuse a caller-supplied
  noncanonical ID.
- **Evidence-pack location:** This file

## Implemented contract

1. The v1 definition requires slot ID, V1–V9 product, surface, content kind,
   unique modalities, canonical BCP 47 locale, audience policy, positive
   quantity, nullable freshness SLO, source/autonomy policies, pipeline, gate
   suite, rights policy, explicit accessibility requirements, placement,
   versioned owner record, risk tier, and legacy aliases. Every object boundary
   is strict.
2. The lifecycle exposes exactly `missing`, `planned`, `running`, `review`,
   `released`, `stale`, `invalidated`, `blocked`, and `withdrawn`, with required
   active-artifact, latest-run, blocker, next-required, and evidence-freshness
   fields. State invariants reject missing blockers, running/review without a
   run, released/stale quantity deficits, released stale evidence, unscheduled
   expiring releases, and active artifacts on missing/invalidated/withdrawn
   slots.
3. `deriveContentSlotId` hashes the canonical
   product/surface/kind/locale/audience/placement tuple into a stable 160-bit
   `cslot_` ID. Identity versioning is independent from record-schema versioning
   so ordinary future schema migrations do not churn IDs.
4. Ownership imports are idempotent at the same revision, reject conflicting or
   stale revisions, advance exactly one revision at a time, require evidence for
   owner/escalation changes, and reject revision churn without a material
   change.
5. The v0 upcaster requires explicit migration policy rather than inventing
   owner, risk, accessibility, or evidence-freshness defaults. It derives the v1
   ID, retains the legacy alias, validates the current aggregate, is repeatable,
   accepts already-current v1 idempotently, and rejects unknown versions and
   unknown weakening fields.

## Publication identity

- **Implementation source commit:** `2ada10c7b9831b18d9217bc64a15352e3e254a3f`
- **Post-merge authenticated-record source:**
  `6f806b6f6cef12d42f4597518ebbe878699504b8`
- **Canonical verification-record publication commit:**
  `16ac905031fb99d65e8725be87c86fc0508c15c2`
- **Verification-record interval:** `2026-07-15T18:04:43.548Z` through
  `2026-07-15T18:09:29.106Z`
- **Host/runtime:** `dev-server`, Linux `6.8.0-110-generic` x64, Node.js
  `v24.14.1`, pnpm `10.25.0`, Vitest `4.1.7`, git `2.43.0`
- **Provider/model gates:** OpenAI, Anthropic, RunPod, and database credentials
  were absent. This schema task makes no provider, placement, or release claim.

| Artifact                                                   | SHA-256                                                            |
| ---------------------------------------------------------- | ------------------------------------------------------------------ |
| `libs/contracts/src/content-coverage/content-slot.ts`      | `abf2498b80ba23834cdc3bfb829ec2531c5571f04a9a1ca83012aa2c4073f2ab` |
| `libs/contracts/src/content-coverage/content-slot.spec.ts` | `0c08f63eef3e135e02897a6c3ff2267cd2bc07f3dc388fa75ba95e2afdf9f1f7` |
| `libs/contracts/src/content-coverage/index.ts`             | `3cc9812a52d7537dcec25acefd307a2ffb6d116648f0c9595f7e39e221f4123d` |
| `.github/workflows/content-slot-contract.yml`              | `e23bec6c096bc5138a5f5938a45e553e40a7d0620f1f24668afe817eddefe4da` |
| `evidence/v1-v9/AUD-003/verification-record.json`          | `533efd50e0f679ed7ede7239a5c8108f431f77b41f2ce17a3ba7526e29cc793d` |

## Automated and adversarial verification

- `pnpm verify:contracts:content-slot` passed targeted ESLint with
  `--no-ignore --max-warnings=0`, 17/17 Vitest cases, and the contracts source
  typecheck. The cases exercise all required fields, all nine states, strict
  nested boundaries, locale/reference uniqueness, accessibility waiver rules,
  state/evidence/quantity/freshness invariants, deterministic IDs, conflicting
  ownership, revision jumps, explicit migration policy, idempotent v0/v1
  processing, and unsupported versions.
- `NX_DAEMON=false pnpm exec nx run @oshun/contracts:build --parallel=1` passed.
  A `tsx` package probe imported both `@oshun/contracts` and
  `@oshun/contracts/content-coverage` and proved the root namespace, flat
  exports, subpath version, and exact nine-state enum agree.
- The resource-bounded package run excluding the two independently failing
  baseline files passed **153/153 files and 3,825/3,825 tests** with
  `--maxWorkers=1`. The unfiltered diagnostic identified only the existing
  `contract-prisma-alignment.test.ts` manifest miss and eight existing Prisma
  CLI-diff failures. `tsconfig.spec.json` likewise retains 47 lines of unrelated
  pre-existing spec errors; neither diagnostic names `content-coverage` or
  `content-slot`.
- `pnpm verify:audit:capability-truth` passed 7/7 tests and all five generated
  views. `pnpm verify:audit:review-reconciliation` passed 7/7;
  `pnpm verify:audit:test-coherence` passed 6/6; and
  `pnpm verify:audit:machine-adjacent-docs` passed 3/3.
- `pnpm generate:audit:v1-v9-verification-record` ran sequentially with
  `--maxWorkers=1` from clean source `6f806b6f6c…`: V1 recorded 248/248,
  289/289, 138/138, and 40/40; V8 recorded 351/351; V9 recorded 176/176. The
  known V8 generated-contract drift remained measured red at exit 1.
- `pnpm verify:audit:v1-v9-verification-record` passed 8/8 adversarial checks
  and record/table/command/boundary/log/source coherence. The record run was
  polled at or below 30-second intervals; available memory remained about 11–12
  GiB and no safety threshold was approached.
- Pre-commit linting, formatting, asset locking, stub scanning, and V5 document
  validation passed. Its inferred root-contract typecheck expanded into an
  unapproved multi-project broad gate, so it was stopped under the 15 GiB host
  policy and bypassed with `SKIP_TYPECHECK=1`; the focused source typecheck,
  package build, package tests, and post-merge focused gate remained green.

## Negative proof and claim boundary

- Deliberately malformed fixtures fail for missing fields; unknown waiver or
  force-release fields; noncanonical locale; duplicate refs/modalities;
  unsupported accessibility waivers; blocked-without-blocker; released without
  fresh/complete/scheduled evidence; active invalidated artifacts; mismatched
  IDs; stale/conflicting ownership; missing migration policy; and unknown schema
  versions.
- This package does not claim a populated coverage registry, dependency graph,
  importer, API, operator UI, placement controller, provider run, released
  artifact, localization coverage, or production rollout. Those remain open in
  COV-002 through COV-005 and the product slices.
- No user-facing UI, mobile flow, media artifact, persistence schema, provider
  request, or Unreal asset changed. Playwright/mobile automation, accessibility
  scans, screenshots, media probes, database migration validation, and Unreal
  execution are therefore inapplicable to COV-001.
- Policy does not require a human signature for this schema-only change. No
  human approval is manufactured or implied; future owner changes must carry
  their own stable change-evidence reference.
