docs: skill inventory and classification matrix for all 38 upstream skills

TODO.md

@@ -13,8 +13,9 @@ State: `.opencode/state/status.md`. Design: `docs/architecture.md`.
 - [x] Gitea Actions design (§6)
 - [x] Testing strategy (§7)
 - [x] Phased plan (§9)
-- [ ] Skill inventory + classification matrix → docs/skill-mapping.md (subagent analysis in flight)
-- [ ] Commit phase-0 docs (`docs: phase 0 analysis and migration plan`)
+- [x] Skill inventory + classification matrix → docs/skill-mapping.md +
+      .opencode/state/classification.json (38 skills analyzed from SKILL.md sources)
+- [x] Commit phase-0 docs
 
 ## Phase 1 — plugin package
 
@@ -36,10 +37,10 @@ State: `.opencode/state/status.md`. Design: `docs/architecture.md`.
 
 ## Phase 2 — conversion
 
-- [ ] 2.1 deep per-skill analysis → .opencode/state/classification.json + docs/skill-mapping.md
-      (classification.json becomes generator input; mapping doc generated from it)
+- [x] 2.1 deep per-skill analysis → .opencode/state/classification.json + docs/skill-mapping.md
+      (done in phase 0; generator must regenerate mapping doc from classification.json — keep in 1.3 scope)
 - [ ] 2.2 guardrails plugin hook (opt-in, tool.hook execute.before on shell) + tests
-- [ ] 2.3 pocock-reviewer: implement or document rejection in decisions.md
+- [ ] 2.3 pocock-reviewer: CONFIRMED by analysis (D8) — implement template + skill-permission scoping
 - [ ] 2.4 orchestrator V2 flow wiring (background-subagent status in V2? batch fallback; worktree
       permission guidance; slash UX for user-invoked skills)
 - [ ] 2.5 per-skill regression checks + manual orchestration smoke (token-spending, documented)