# OSHUN Metis V1 Scorecard

Status: Active

Date: April 9, 2026

Owners: OSHUN Product, OSHUN Engineering, OSHUN Platform, OSHUN Safety

Related:

- `docs/releases/v1/reports/metis-domain-gap-analysis-and-full-integration-report-2026-03-23.md`
- `docs/releases/v1/reports/metis-subsystem-ownership-matrix.md`
- `docs/releases/v1/specs/metis-mobile-strategy.md`
- `docs/releases/v1/runbooks/metis-operator-workflows-runbook.md`
- `docs/releases/v1/specs/metis-generation-and-tutoring-slos.md`
- `docs/releases/v1/specs/metis-media-runtime-tiers.md`
- `docs/releases/v1/specs/metis-persistence-migration-backfill-plan.md`
- `docs/releases/v1/specs/metis-learning-source-bundle-contract.md`
- `docs/releases/v1/specs/metis-grounding-pack-contract.md`
- `docs/releases/v1/specs/metis-course-build-contract.md`
- `docs/releases/v1/specs/metis-lesson-asset-bundle-contract.md`
- `docs/releases/v1/specs/metis-publication-package-contract.md`
- `docs/releases/v1/specs/metis-avatar-teacher-profile-contract.md`
- `docs/releases/v1/specs/metis-voice-likeness-consent-record-contract.md`
- `docs/domains/metis/METIS_KNOWLEDGE_GRAPH_AND_RETRIEVAL_ALGORITHM_STATUS.md`
- `services/metis/openapi/metis.openapi.json`
- `libs/openapi/src/generated/metis.ts`
- `docs/releases/v1/specs/program-charter.md`
- `docs/releases/v1/specs/studio-launch-readiness-governance-dependency-and-data-contract-map.md`
- `docs/releases/v1/specs/studio-launch-readiness-governance-user-journeys-and-edge-cases.md`
- `docs/releases/v1/specs/todos.md`

## 1. Purpose

This document is the canonical Metis readiness scorecard for OSHUN V1.

It consolidates the learner, admin, platform, governance, and release-gate
dimensions that must be green before Metis can be treated as V1-complete.

This is not a feature inventory and not a marketing status page. It is a launch
control artifact that should stay aligned with verified runtime truth.

## 2. Status Scale

- `Green`: launch-grade, verified in code/runtime, and covered by applicable
  automation and operational evidence
- `Amber`: meaningful foundation exists, but launch criteria are still unmet
- `Red`: launch-blocking capability is missing, mock-driven, simulated, or
  insufficiently governed

## 3. Current Snapshot

| Scorecard lane            | Current status | Why                                                                                                                                                                                       |
| ------------------------- | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Learner completeness      | `Red`          | The shell-level Metis entry points now exist, but `apps/metis/web` remains materially mock-backed across auth, browse, progress, tutoring, assessment, and authoring flows.               |
| Admin completeness        | `Red`          | The Metis admin surface is still largely mock-driven, and the required review, moderation, audit, support, and urgent-queue paths are not yet live and verified.                          |
| Platform completeness     | `Amber`        | Real libraries, gateway, worker, and service scaffolding exist, but the platform still lacks complete contract publication, service-test depth, and end-to-end API/runtime proof.         |
| Governance completeness   | `Red`          | Ownership is now explicit, but launch-grade rights/integrity workflows, approval controls, and remaining governance gates are still incomplete even with operator runbooks now published. |
| Release-gate completeness | `Amber`        | The Studio readiness model already requires a dedicated Metis lane, but Metis-specific quality, rights, safety, cost, pedagogy, and provenance gates are not yet all operationally green. |

## 4. Learner Completeness

Learner completeness is green only when all of the following are true:

| Dimension                                   | Current status | Green criteria                                                                                              | Primary backlog drivers                                       |
| ------------------------------------------- | -------------- | ----------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------- |
| Live learner auth and session truth         | `Red`          | learner identity, session hydration, and continuity are API-backed rather than mock-backed                  | `V1-MET-015`                                                  |
| Live course browse, detail, and progress    | `Red`          | browse, detail, resume, and progress state are live and durable                                             | `V1-MET-016`, `V1-MET-017`, `V1-MET-024`, `V1-MET-027`        |
| Live tutoring and assessment flows          | `Red`          | tutoring and assessments are session-backed, grounded where required, and recoverable after interruption    | `V1-MET-018`, `V1-MET-019`, `V1-MET-069` through `V1-MET-085` |
| Shell and mobile continuity                 | `Amber`        | shell entry, mobile strategy, resume paths, reminders, and narrow-width behavior are coherent and verified  | `V1-MET-021` through `V1-MET-028`                             |
| Trust, accessibility, and degraded-state UX | `Red`          | grounded evidence, accessibility, offline/degraded, and export/share behavior exist and are learner-legible | `V1-MET-025`, `V1-MET-026`, `V1-MET-029`, `V1-MET-030`        |
| Learner automation and performance          | `Red`          | Playwright coverage and route-level performance/PWA checks pass for critical learner journeys               | `V1-MET-031`, `V1-MET-032`, `V1-MET-107`                      |

## 5. Admin Completeness

Admin completeness is green only when all of the following are true:

| Dimension                                  | Current status | Green criteria                                                                                                                  | Primary backlog drivers                  |
| ------------------------------------------ | -------------- | ------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------- |
| Live admin dashboards and data truth       | `Red`          | dashboards, users, courses, and flagged content are live, not mock-backed                                                       | `V1-MET-033`                             |
| Review and approval queues                 | `Red`          | source bundles, grounded outputs, tutor personas, prompts, avatars, voice, and rights all have operator-grade review queues     | `V1-MET-034` through `V1-MET-039`        |
| Monitoring, incidents, and support tooling | `Red`          | latency/grounding/failure panels, learner complaint tooling, and incident workflows are live and actionable                     | `V1-MET-040`, `V1-MET-043`, `V1-MET-044` |
| Audit, permissions, and user actions       | `Red`          | Metis admin actions have durable audit trails, user-management controls, and contract/permission verification                   | `V1-MET-041`, `V1-MET-042`, `V1-MET-047` |
| Admin automation and mobile urgent queues  | `Red`          | browser automation exists for review/moderation/monitoring, and admin mobile escalation coverage exists for urgent Metis queues | `V1-MET-045`, `V1-MET-046`               |
| Operator documentation and training        | `Green`        | runbooks, training, and operator-facing documentation are published and current                                                 | `V1-MET-048`                             |

## 6. Platform Completeness

Platform completeness is green only when all of the following are true:

| Dimension                                   | Current status | Green criteria                                                                                                         | Primary backlog drivers                                                                                                 |
| ------------------------------------------- | -------------- | ---------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------- |
| Canonical contracts                         | `Amber`        | learner, admin, media, and workflow contracts are published and versioned in canonical Metis OpenAPI/contract surfaces | `V1-MET-049` through `V1-MET-053`, `V1-MET-101`                                                                         |
| API-backed runtime truth                    | `Amber`        | ingest, retrieval, generation, tutoring, assessment, and publication paths execute against live backend contracts      | `V1-MET-054` through `V1-MET-067`, `V1-MET-102`, `V1-MET-103`                                                           |
| Service and persistence reliability         | `Amber`        | durable state, migrations, job reliability, and dependency behavior are verified for service-critical paths            | `V1-MET-104`, `V1-MET-105`, `V1-MET-109`; migration baseline in `OSHUN_V1_METIS_PERSISTENCE_MIGRATION_BACKFILL_PLAN.md` |
| Interoperability and export readiness       | `Amber`        | LMS/import/export promises are only exposed where implemented, tested, and operationally supported                     | `V1-MET-064`, `V1-MET-099`, `V1-MET-157` through `V1-MET-195`                                                           |
| Platform observability and rollback posture | `Amber`        | build-job, tutor-session, publication, and interoperability telemetry are wired to rollback-aware operational signals  | `V1-MET-109`, `V1-MET-196`                                                                                              |

## 7. Governance Completeness

Governance completeness is green only when all of the following are true:

| Dimension                              | Current status | Green criteria                                                                                                      | Primary backlog drivers                                              |
| -------------------------------------- | -------------- | ------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------- |
| Ownership and subsystem boundaries     | `Amber`        | Metis-to-substrate ownership is explicit and current in canonical docs                                              | `V1-MET-012`                                                         |
| Source, rights, and integrity review   | `Red`          | grounded source approval, licensing review, originality checks, moderation, and appeals are live and auditable      | `V1-MET-039`, `V1-MET-074`, `V1-MET-081`, `V1-MET-114`, `V1-MET-145` |
| Tutor and pedagogy governance          | `Red`          | persona, pedagogy mode, escalation, and sensitive-teaching controls are reviewable and approval-backed              | `V1-MET-035`, `V1-MET-073`, `V1-MET-076`, `V1-MET-082`               |
| Avatar, voice, and consent governance  | `Red`          | avatar teacher profiles, likeness consent, safety, fallback rules, and approvals are operational                    | `V1-MET-036`, `V1-MET-090`, `V1-MET-095`, `V1-MET-098`               |
| Documentation and operational currency | `Amber`        | scorecards, ownership, runbooks, operator docs, and truth-in-documentation stay aligned with verified runtime state | `V1-MET-012`, `V1-MET-013`, `V1-MET-048`, `V1-MET-197`               |

## 8. Release-Gate Completeness

Release-gate completeness is green only when all of the following are true:

| Dimension                                         | Current status | Green criteria                                                                                                             | Primary backlog drivers                                                            |
| ------------------------------------------------- | -------------- | -------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------- |
| Dedicated Metis lane in readiness governance      | `Amber`        | Metis learner/tutoring/admin scope always generates a dedicated scorecard lane and blocker queue                           | `V1-GOV-027`, `V1-MET-013`                                                         |
| Quality, rights, safety, cost, and pedagogy gates | `Amber`        | launch gates evaluate educational quality, rights, safety, cost, and pedagogy rather than generic uptime only              | `V1-MET-040`, `V1-MET-082`, `V1-MET-096`, `V1-MET-109`, `V1-MET-110`, `V1-MET-116` |
| Provenance and confidence evidence                | `Amber`        | release decisions carry claim-level provenance and confidence evidence for grounded educational outputs                    | `V1-MET-063`, `V1-MET-068`, `V1-MET-116`                                           |
| Benchmark and adversarial evidence                | `Red`          | benchmark suites, red-team cases, prompt-injection resistance, and governed holdout evaluation are attached to gate review | `V1-MET-068`, `V1-MET-079`, `V1-MET-081`, `V1-MET-082`, `V1-MET-131`, `V1-MET-196` |
| Exception handling and rollback triggers          | `Amber`        | open blockers, exceptions, and rollback triggers are explicit, time-bounded, and audit-backed for Metis scope              | `V1-MET-044`, `V1-MET-109`, `V1-MET-196`                                           |

## 9. Green-Only Exit Rule

Metis should be considered V1-complete only when:

- learner completeness is green
- admin completeness is green
- platform completeness is green
- governance completeness is green
- release-gate completeness is green or has only time-bounded, approval-backed
  exceptions consistent with the OSHUN V1 charter

If any lane is red, Metis is launch-blocking.

If any lane is amber, Metis is not yet complete even if portions of the stack
look production-adjacent.

## 10. Full Integration Hold Rule

Metis must not be marked fully integrated into OSHUN V1 while any launch-
critical learner or admin path remains mock-driven, stub-dependent, simulated,
or verified only against non-canonical local data.

This rule is fail-closed.

If any of the following are true, the canonical answer is still "not fully
integrated":

- `apps/metis/web` still depends on material mock-backed learner flows
- `apps/metis/admin` still depends on material mock-backed operator flows
- scorecard lanes remain `Red`
- scorecard lanes remain `Amber` without time-bounded, approval-backed
  exceptions
- applicable automated verification for learner, admin, API, release-gate, or
  operational paths is not passing

Current launch hold evidence:

- learner completeness remains `Red`
- admin completeness remains `Red`
- governance completeness remains `Red`
- `apps/metis/web/src/lib/mock-data.ts` and
  `apps/metis/admin/src/lib/mock-data.ts` still exist as material truth sources
  for current product surfaces
- multiple Metis browser suites still rely on deterministic fixture layers
  rather than fully live end-to-end runtime truth

Metis may only be marked fully integrated after the scorecard itself shows all
required lanes green and the linked verification evidence is current.
