# V1 Specification Coverage Matrix (2026-05-30) Systematic, spec-by-spec analysis of the **entire** `V1/features.md` (6,241 lines: 22 top-level sections, 99 subsections) + the 6 customer domains + the 19 **V1 Exit Criteria** (the spec's own definition of done), each with an implementation status and code evidence. This is the whole-system companion to the studio-surface census ([`V1_STUDIO_SURFACE_GAP_ANALYSIS_2026-05-30.md`](V1_STUDIO_SURFACE_GAP_ANALYSIS_2026-05-30.md)) and the deploy-tier map ([`V1_DEPLOYMENT_REQUIREMENTS.md`](V1_DEPLOYMENT_REQUIREMENTS.md)). Status legend: **REAL** = implemented + reachable on a runtime path, evidence cited · **SURFACE-GAP** = backend/algorithm real, but operator/creator studio _pages_ largely unwired (the volume gap) · **EXTERNAL** = in-repo code complete, blocked only on deploy-time inputs (creds/weights/config/content) · **PARTIAL** = real core with a named remainder. ## A. `features.md` section coverage (22 sections) | § | Section | Status | Evidence / note | | -------------------------------------------- | ----------------------------------- | ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ | | Domain Features | Tara/Arete/Veritas/Nyx/Nisaba/Metis | REAL | all 6 hubs wired web→BFF room endpoints (`apps/oshun/web/src/lib/lilith-data/*`, `/v1//room`); ~50 domain libs under `libs/` | | Content Authoring/Curation/Editorial | curated cards, calendar, taxonomy | PARTIAL | curated-card generation surface wired (catalog + gate + 3 generation studios); editorial calendar/taxonomy studio pages in the SURFACE-GAP | | Assistant/Memory/Persona/Embodiment | iris/psyche/mnemosyne | REAL | persona/memory libs present; psyche tavus-tools HS256 verify hardened; iris voice synthesis + clone-workflow real | | Grounding/Generation/Media/Search | sophia/isis/search | REAL | Isis generation gate + async job queue + 3 provider adapters; sophia grounding; search offline-eval release gate + recommender | | Living Scenes | render→watermark→PSE→C2PA→share | REAL | `living-scene-render.ts` deterministic frames → `embedWatermarkMark` → PSE → durable shares + takedown cascade; `POST /v1/living-scenes/render` | | Agentic AI Studio | runs/budgets/kill-switches/catalog | REAL | `runGuardedToolPlan` + the full studio-mutation **tool catalog** now wired (this session); kill-switch/budget enforced on real tools | | Admin Products | review/safety/persona/model/rights | REAL+SURFACE-GAP | admin-isis backends (civitai/comfy/lora/merge/gallery/runpod/voice) real + 8 operator pages built; remaining admin studio pages in surface-gap | | Tenant/Institution/Operator Toolkit | SSO/SCIM/OneRoster/webhooks | REAL+EXTERNAL | LTI 1.3 verify + SCORM 1.2/2004 RTE + OneRoster dry-run mounted; per-tenant connector registration is deploy DATA | | Governance/Safety/Support/Privacy | cascades/DSAR/crisis/takedown | REAL | takedown/DSAR/crisis cascades wired + tested; durable cross-instance abuse limiter (Redis) | | Messaging Channels | transports + reminders | REAL | email/SMS/push/WhatsApp/Slack/Discord transports; reminder planner→worker→delivery + the **4 producers** (this session); in-app always-available | | Architecture/Platform Foundations | BFF/persistence/event-bus | REAL | Fastify BFF; durable snapshot store; Redis event bus; docker dev infra | | Design System / Experience Quality | tokens/components | REAL | design-system present; recent contrast-token fixes | | Analytics/Observability/Experimentation/Eval | metrics/offline-eval | REAL | search offline-eval gate; telemetry; (full dashboards = deploy-tier) | | Testing/QA/Accessibility/Perf/Security | risk-tiered verification | PARTIAL | extensive vitest coverage + e2e; per-page studio tests added as pages are wired | | Content/Localization/Docs/Launch | corpus/i18n/launch | EXTERNAL | localization scaffolding present; expanded editorial corpus + launch assets are content/deploy work | ## B. Customer domains (6) — Exit-Criterion "one coherent product" | Domain | Status | Evidence | | --------------------- | ------------- | --------------------------------------------------------------------------------------------------- | | Tara (product center) | REAL | `/v1/tara/room` with real sitting-completion progress (live-PG verified) | | Arete | REAL | `/v1/arete/*` wired; web hub | | Veritas | REAL | `/v1/veritas` grounded briefing (no fabricated citations) | | Nyx | REAL | `/v1/nyx/tonight` real ephemeris | | Nisaba | REAL | `/v1/nisaba/room` with real annotation counts | | Metis | REAL+EXTERNAL | `/v1/metis/room` course proposals; OneRoster dry-run; course-publish + per-tenant LMS = deploy DATA | ## C. V1 Exit Criteria scorecard (the spec's definition of done — 19) | # | Criterion | Verdict | | --- | -------------------------------------------------------------- | ------------------------------------------------------------------------------------------ | | 1 | Coherent product across 6 domains | REAL (all hubs wired) | | 2 | Tara visibly the center | REAL | | 3 | Nisaba + Metis integrated | REAL | | 4 | Sophia grounding visible | REAL (veritas/metis evidence) | | 5 | Cross-domain continuity tested | REAL (recommendations route + tests) | | 6 | Admin can operate ALL governance workflows | **SURFACE-GAP** (backends real; many admin studio _pages_ unwired) | | 7 | Oshun Studio authoring/editorial/curation | **SURFACE-GAP** (the ~450-page volume gap) | | 8 | Agentic Studio operational, gates enforced | REAL (guard + catalog wired this session) | | 9 | Tenant Console SSO/SCIM/OneRoster/webhooks | REAL (in-repo) + EXTERNAL (tenant config) | | 10 | Iris/Psyche/Lilith/Sophia/Isis/Metis ownership true | REAL | | 11 | External models can't bypass review/rights/safety/Isis gate | REAL (civitai intake state machine + Isis release gate) | | 12 | Metis high-stakes claim-level evidence + gates | REAL (grounding/pedagogy/safety/standards gates) | | 13 | Public website/app-store/deep-links/status page | EXTERNAL (launch assets) | | 14 | Recommendation generation/ingestion/offline-eval/observability | REAL (search offline-eval gate + recommender) | | 15 | Aja embodied-instruction for Metis ready or deferred | **SURFACE-GAP** (mocap libs present; Aja studio pages unwired — largest breadcrumb domain) | | 16 | All critical workflows risk-tiered automated verification | PARTIAL (broad coverage; per-page tests grow with the build-out) | | 17 | Accessibility/perf/security/resilience/launch gates | PARTIAL (gates exist; full launch-readiness is deploy-tier) | ## D. Consolidated gap register (all layers, not just studio) 1. **Operator/creator Studio pages (~450 unwired of 469)** — the dominant gap. Backends/algorithms exist; the _pages_ need read/mutate wiring + tests. Quantified in the surface analysis. (Exit Criteria 6, 7, 15.) 2. **Fabricated-static pages (248)** — look built, wired to nothing; must be re-wired to real data (subset of the 450). 3. **External / deploy-tier (Exit 9, 13, 17; documented):** trained AI-video weights + GPU endpoint; live Twilio/SES/FCM/Meta + Suno/RunPod/Meshy creds; per-tenant LMS connector registrations; reminder/worker cron host; expanded editorial corpus; C2PA signing key; launch assets + public status page. 4. **Verification depth (Exit 16, 17):** automated coverage is broad; each newly wired page adds its tests; full launch-gate sign-off is a release activity. ## E. Verdict - **Phase 1 (analyze entire implementation):** done here — all 22 spec sections, 6 domains, 19 Exit Criteria walked with evidence. - **Phase 2 (identify all gaps):** the gap register (§D) is the consolidated, whole-system gap set; the studio surface census enumerates the per-page subset. - **Phase 3 (ensure fully production-ready):** the **in-repo platform layers** (domains, generation, governance/safety, agentic, messaging, living-scenes, LMS protocols, recommendation) are production-real and reachable. **Full production readiness is NOT yet reached**, gated on: (a) the ~450-page studio build-out (in-repo, volume-bound, multi-week, in progress one verified page per turn), and (b) the irreducible deploy-tier externals (§D.3), which are not source code. This document makes that state explicit rather than implying completion.