> For the complete documentation index, see [llms.txt](https://ai-os-and-trend-finder.gitbook.io/ai-os-and-trend-finder-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ai-os-and-trend-finder.gitbook.io/ai-os-and-trend-finder-docs/docs/docs-audit.md). # Documentation Audit Strategy **DO ONLY ONE GROUP PER SESSION** ## Group Checklist - DO ONLY ONE GROUP PER SESSION * [x] Group 0 - Audit Control Plane * [x] Group 1 - Root Project Contract * [x] Group 2 - Product Source Of Truth * [x] Group 3 - Setup, Development, Environments, Deployment * [x] Group 4 - AI Runtime, Commands, Data, Agents, Skills * [x] Group 5 - Extension Overview And Trend Finder Concepts * [x] Group 6 - Trend Finder Scoring, Sources, Creator Lens * [x] Group 7 - Runbooks And Operations * [x] Group 8 - Hackathon, Public Demo, And Research Notes * [x] Group 9 - ADRs, API, And Planning Records * [x] Group 10 - Directory READMEs * [x] Group 11 - Local Skill Docs * [x] Group 12 - Example And Imported Reference Docs * [x] Group 13 - Generated Or Ignored Audit Artifacts * [ ] **Strategy date**: 2026-05-26 * [ ] **Project**: AI OS * [ ] **Mode**: Full documentation audit planning * [ ] **Audit status**: Complete as of the 2026-05-26 final wrap * [ ] **Purpose**: Divide every project documentation surface into small, context-safe audit groups so each pass can verify accuracy thoroughly. ## Operating Rules CORE RULE: **DO ONLY ONE GROUP PER SESSION** 1. Audit in small groups. Target 4-8 active files per pass, or fewer when files are long, dense, or source-of-truth documents. 2. Keep one source of truth. Update duplicate docs by linking to the owner doc instead of repeating large sections. 3. Separate implemented behavior from planned, deferred, historical, imported, or generated behavior. 4. Preserve identity boundaries: use AI OS for the host shell and platform docs; use Trend Finder for extension routes, collectors, public source evidence, scores, creator angles, watchlists, hackathon docs, and demos. 5. Preserve privacy boundaries. Do not inspect or document private generated data contents such as `src/data/live-data.json`, private logs, local source directories, API keys, or `.env.local`. 6. Treat generated or ignored docs as separate from active docs. They may need disposition notes instead of content updates. 7. After each group, record changed files, stale claims removed, validation performed, and any follow-up group that must consume the result. ## Inventory Command Use this before each full audit pass so newly added docs are not missed: ```bash find . \ -path ./node_modules -prune -o \ -path ./.git -prune -o \ -path ./dist -prune -o \ -path ./build -prune -o \ -path ./coverage -prune -o \ -type f \( -iname '*.md' -o -name 'README_*' -o -name 'LICENSE' -o -name 'CODEOWNERS' \) \ -printf '%P\n' | sort ``` Current snapshot: 79 document-like files outside ignored dependency/build directories, plus root symlink aliases `CLAUDE.md -> AGENTS.md` and `GEMINI.md -> AGENTS.md`. ## Audit Session Log ### 2026-07-03 - Full Access Documentation Default Update Completed: * Added the project-wide local access default: AI OS operator goals default to full local access, write access, and edit access. * Updated active docs so manual admin gates, legacy limited-access route names, metadata-only scanners, unsupported preflights, and demo states are described as current implementation constraints or explicit exceptions, not desired defaults. * Recorded the delivery bar: shipped features require real execution, visible results, recovery paths, and tests. * Preserved public demo, source-compliance, credential, privacy, and external dependency exceptions as explicit boundaries. Changed files: * `AGENTS.md` * `README.md` * `.spec_system/PRD/PRD.md` * `.spec_system/CONSIDERATIONS.md` * `.spec_system/SECURITY-COMPLIANCE.md` * `.spec_system/docs-audit.md` * `docs/ARCHITECTURE.md` * `docs/api/README_api.md` * `docs/agent-pages.md` * `docs/commands.md` * `docs/development.md` * `docs/environments.md` * `docs/intelligence-view.md` * `docs/local-agent-contract.md` * `docs/local-voice-setup.md` * `.spec_system/PRD/phase_41/PRD_phase_41.md` * `.spec_system/PRD/phase_41/session_*.md` * `docs/skill-system.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Targeted `rg` scans for active documentation access-default language. * Targeted review of root README, PRD, architecture, development, environment, local API, agent page, voice, Intelligence, local-agent, command, skill, and spec-system memory/security docs. Remaining: * Product implementation still needs the planned all-access sessions in `.spec_system/PRD/phase_41/` before docs can claim every Hermes, Knowledge Graph, Voice, Intelligence, OpenClaw, and Claude Code path is write-ready by default in running code. ### 2026-07-03 - Phase 40 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 40 Claude OS v2.10.1 Semantic Port release. * Verified Phase 40 delivered Hermes model/provider contracts, shared redaction, chat overrides, command and MoA admin endpoints, connection probes, catalog/context metadata, model intelligence, pricing provenance, provider assets, chat selector/context meter, command UX, Ministry builder and save behavior, voice parity, docs closeout, and final validation. * Verified current README, API notes, local voice setup, Intelligence, phase closeout, environment, deployment, and security docs describe shipped AI OS behavior and keep live-provider, local-admin, and production-infra proof limits explicit. * Updated the architecture endpoint inventory for the Phase 40 Hermes model intelligence, command, MoA save, mission commit/set-active, and voice launch routes. * Replaced the Phase 39 spec-system documentation audit with the Phase 40 audit report. Changed files: * `docs/ARCHITECTURE.md` * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/PRD/PRD.md` * Review of `.spec_system/archive/phases/phase_40/` * Review of Phase 40 implementation notes, validation reports, security-compliance reports, implementation summaries, final validation log, and final handoff for Sessions 01-18 * Review of `package.json`, root README, architecture, development, environments, deployment, API notes, phase closeout, local voice setup, Intelligence view, required root docs, ADRs, runbooks, and workflow files * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped endpoint, stale-claim, ASCII/LF, and `git diff --check` validation for changed documentation files Remaining: * No Phase 40 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` has no remaining unfinished phase after Phase 40, and the analyzer reports no candidate sessions, so the workflow is complete. * Environment-limited proof remains intentionally documented as a caveat for live Realtime voice, local-admin MoA save success, production Worker health, Cloudflare WAF ruleset verification, and GitHub Actions deploy execution. * Legacy nested `README.md` paths remain in generated, public asset, demo, imported example, and external cache scopes. This phase-focused pass did not rename them because active and historical links still target those document maps. ### 2026-07-01 - Phase 39 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 39 AI Rogue Level Authoring Infrastructure release. * Verified Phase 39 delivered registry-backed AI Rogue level specs, depth resolver migration, save-ID parity, a fourth authored floor, enemy metadata, derived asset and audio gates, boss/finale contracts, the Insight Beetle real-content path, and closeout validation/docs hardening. * Verified AI Rogue docs already describe the four-level authored main run, `runtime/content/` owner, Firewall Gauntlet, Insight Beetle, Kernel Sentinel finale metadata, bundled audio, save schema version 1, and no-new-media closeout. * Fixed the root README AI Rogue docs link to the active `README_ai-rogue.md` file. * Replaced the Phase 38 spec-system documentation audit with the Phase 39 audit report. Changed files: * `README.md` * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of `.spec_system/archive/phases/phase_39/` * Review of Phase 39 implementation notes, validation reports, security reports, and implementation summaries for Sessions 01-08 * Review of `package.json`, root README, architecture, AI Rogue docs, testing, changelog, API notes, required root docs, ADRs, and runbooks * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped link, stale-claim, ASCII/LF, and `git diff --check` validation for changed documentation files Remaining: * No Phase 39 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` has no remaining unfinished phase after Phase 39, so the workflow is complete. * Legacy nested `README.md` paths remain in generated, public asset, demo, and imported example scopes. This phase-focused pass did not rename them because active and historical links still target those document maps. * Future AI Rogue content classes, remote loading, hosted writes, collectors, workers, WebGPU-specific behavior, saved audio state, map-editor dependency, and rejected-art reuse remain explicitly out of scope pending fresh review. ### 2026-06-30 - Phase 38 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 38 Claude OS v2.8.1 Semantic Port release. * Verified Phase 38 delivered Tier 0 parity fixes, shared platform helpers, Windows-aware aggregate and Dream behavior, runtime bridge hardening, policy and model catalog updates, Dream engine integration, local voice broker support, the Hermes Intelligence portal, and final hunk reconciliation. * Verified current docs already cover Session 08 voice broker ownership, Session 09 Intelligence portal ownership, env-only provider keys, and the unverified live-provider caveat. * Verified root, architecture, onboarding, development, environment, Dream, voice, Intelligence, changelog, API, runbook, ADR, and required root docs are present and current for Phase 38 behavior. * Replaced the Phase 37 spec-system documentation audit with the Phase 38 audit report. Changed files: * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of `.spec_system/archive/phases/phase_38/` * Review of Phase 38 implementation notes, validation reports, security reports, and implementation summaries for Sessions 01-10 * Review of `package.json`, root README, architecture, onboarding, development, environments, AI OS Dream runbook, local voice setup, Intelligence view, changelog, API notes, required root docs, ADRs, and runbooks * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, ASCII/LF, phase current-state claim review, and `git diff --check` validation for changed documentation files Remaining: * No Phase 38 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` has no remaining unfinished phase after Phase 38, so the workflow is complete. * Legacy nested `README.md` paths remain in extension, public asset, generated, and example scopes. This phase-focused pass did not rename them because active and historical links still target those document maps. * Real Windows/macOS scheduler proof and live provider voice proof remain environment-limited validation caveats, not documentation gaps. Current docs avoid claiming those unverified results. ### 2026-06-29 - Phase 37 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 37 AI Rogue Visual Asset Finishing release. * Verified Phase 37 delivered G8 runtime FX, G4 status/equipment icons, G5 boss/final-defense art, G3 sector decals, G6 UI/cinematic panels, and the G7 player-animation rejection decision. * Updated root and architecture summaries so the committed AI Rogue visual atlases are visible from the main project docs. * Updated the active AI Rogue game-feel asset contract with the current 234 gameplay frames, 146 UI frames, G3-G8 frame groups, and stable four-frame player contract. * Replaced the Phase 36 spec-system documentation audit with the Phase 37 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/extensions/ai-rogue/game-feel.md` * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 37 implementation notes, validation reports, security reports, and summaries for Sessions 01-06 * Review of `package.json`, root README, architecture, active AI Rogue docs, crop manifest, runtime atlas metadata, generated preview files, source/test paths, and browser test evidence * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, ASCII/LF, phase visual claim search review, and `git diff --check` validation for changed documentation files Remaining: * No Phase 37 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` has no remaining unfinished phase after Phase 37, so the workflow is complete. * Legacy nested `README.md` paths remain in extension, public asset, generated, and example scopes. This phase-focused pass did not rename them because active and historical links still target those document maps. ### 2026-06-28 - Phase 36 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 36 AI Rogue Audio Asset Finishing release. * Verified Phase 36 delivered typed enemy-family audio metadata, enemy/boss SFX, theme ambience routing and assets, adaptive request ducking, dedicated stingers, provenance coverage, browser proof, and operator-approved manual listening closeout. * Updated root and architecture summaries so committed AI Rogue Web Audio music, theme ambience, SFX, and adaptive stingers are visible from the main project docs. * Verified active AI Rogue game-feel and media-policy docs already record the 11 music/ambience files, 64 SFX/stinger files, provenance coverage, validation evidence, and no widened privacy/runtime boundary. * Replaced the Phase 35 spec-system documentation audit with the Phase 36 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 36 implementation notes, validation reports, security reports, and summaries for Sessions 01-08 * Review of `package.json`, root README, architecture, media-policy, active AI Rogue docs, provenance JSON, audio asset inventory, and browser test evidence * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, ASCII/LF, stale-audio-claim search review, and `git diff --check` validation for changed documentation files; the only stale-audio search hit was a historical/superseded doc-map label. Remaining: * No Phase 36 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` defines Phase 37 as not started, so the next workflow command is `phasebuild`. * Legacy nested `README.md` paths remain in extension, public asset, generated, and example scopes. This phase-focused pass did not rename them because active and historical links still target those document maps. ### 2026-06-27 - Phase 35 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 35 AI Rogue Audit Hardening And Refactor release. * Verified Phase 35 Session 09 synchronized AI Rogue audio, media policy, historical supersession, production enablement, host-LiveData, browser-local state, no-collector, and public-demo no-bridge documentation. * Verified Phase 35 Session 10 recorded Production Go after final app, AI Rogue, Pages, privacy, budget, asset, playthrough, and release-evidence gates. * Updated active root and AI Rogue documentation so the 2026-06-27 final gate, tight bundle-budget caveat, and active-at-cap playthrough caveat are visible. * Fixed active references from the old Phase 35 PRD path to the archived phase artifact path. * Replaced the Phase 34 spec-system documentation audit with the Phase 35 audit report. Changed files: * `README.md` * `.spec_system/PRD/PRD.md` * `docs/extensions/ai-rogue/README_ai-rogue.md` * `docs/extensions/ai-rogue/enablement-decision.md` * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of archived Phase 35 PRD and Session 09/10 implementation, validation, security, and summary artifacts * Review of root, architecture, media-policy, and active AI Rogue docs * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, ASCII/LF, active path-reference search, and `git diff --check` validation for changed documentation files Remaining: * No Phase 35 user-facing repository documentation gaps remain from this pass. Existing historical spec artifacts may still quote pre-archive Phase 35 PRD paths as part of their execution evidence. The active source now points to `.spec_system/archive/phases/phase_35/`. ### 2026-06-26 - Phase 34 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 34 AI Rogue Audit Remediation release. * Verified Phase 34 closed the audited AI Rogue default-enable blockers and AI Rogue is now production-enabled by default in the browser registry. * Updated active root, architecture, conventions, and AI Rogue playtest docs so the Production Go recommendation, current media-policy exception, and latest automated browser/playthrough evidence are discoverable. * Replaced the Phase 33 spec-system documentation audit with the Phase 34 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/CONVENTIONS.md` * `docs/extensions/ai-rogue/playtest-notes.md` * `docs/docs-audit.md` * `.spec_system/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 34 implementation notes and closeout gate evidence * Review of `package.json`, README, architecture, conventions, media policy, AI Rogue active docs, extension runtime docs, and source/test paths named in Phase 34 implementation notes * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, ASCII/LF, active AI Rogue stale-claim searches, and `git diff --check` validation for changed documentation files Remaining: * No Phase 34 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` defines no unfinished phase after Phase 34, so the workflow is complete until a new phase is added to the PRD. ### 2026-06-25 - Phase 33 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 33 Cloudflare Pages Real Product Fixtures release. * Updated active public-demo documentation so the frozen real Trend Finder and Dream Review fixture counts, source commit, capture time, route coverage, scan result, and hosted verification are discoverable. * Renamed the active public-demo README from `demo-website/README.md` to `demo-website/README_demo-website.md` to match the spec-system subdirectory README naming rule, then updated active links. * Documented the isolated static-preview fallback for Wrangler Pages dev when repository Worker configuration interferes with serving `demo-website/dist`. * Replaced the Phase 32 spec-system documentation audit with the Phase 33 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/README_docs.md` * `docs/deployment.md` * `docs/development.md` * `demo-website/README_demo-website.md` * `.spec_system/docs-audit.md` * `docs/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 33 specs, implementation notes, validation reports, security-compliance reports, and deployment verification * Review of `package.json`, `demo-website/snapshot-manifest.json`, `demo-website/public/demo/snapshot-metadata.json`, README, architecture, deployment, development, and documentation index surfaces * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, markdownlint, ASCII/LF, active Markdown link search, and `git diff --check` validation for changed documentation files Remaining: * No Phase 33 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` defines no unfinished phase after Phase 33, so the workflow is complete until a new phase is added to the PRD. ### 2026-06-25 - Phase 32 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 32 AI Rogue Mobile Input Auto-Detect release. * Updated active platform and AI Rogue docs so the shipped raw Auto preference, mounted pointer/hover resolver, concrete `keyboard | compact` runtime boundary, explicit override behavior, and public-demo gameplay coverage are discoverable. * Fixed the active PRD Phase 32 broken link to the retired ongoing-project research note by describing the Session 01 baseline validation evidence directly. * Removed stale no-audio non-goal wording from the touched AI Rogue platform docs because AI Rogue audio is now implemented and documented in the extension docs. * Replaced the Phase 31 spec-system documentation audit with the Phase 32 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/development.md` * `docs/testing.md` * `docs/extensions/ai-rogue/README_ai-rogue.md` * `docs/extensions/ai-rogue/runtime-data-and-enablement.md` * `docs/PRD.md` * `.spec_system/PRD/PRD.md` * `.spec_system/docs-audit.md` * `docs/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 32 specs, implementation notes, validation reports, and implementation summaries * Review of `package.json`, AI Rogue source/test files, Playwright specs, README, architecture, development, testing, and AI Rogue extension docs * Required documentation presence check for root docs, core docs, ADRs, runbooks, and API docs * Scoped Prettier, markdownlint, ASCII/LF, stale-link search, and `git diff --check` validation for changed documentation files Remaining: * No Phase 32 user-facing repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` defines Phase 33 as not started, so the next workflow command is `phasebuild`. ### 2026-06-24 - Phase 31 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 31 Cloudflare Pages Public Demo release. * Verified the public demo uses committed browser-safe fixtures under `demo-website/public/demo/`, generated Pages output under `demo-website/dist/`, and a static-only deployment path separate from the existing Cloudflare Worker deployment. * Updated root, architecture, development, and documentation index surfaces so the public demo quick path, demo scripts, fixture boundary, and route-smoke validation path are discoverable. * Replaced the stale Phase 30 spec-system documentation audit with the Phase 31 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/README_docs.md` * `docs/development.md` * `.spec_system/docs-audit.md` * `docs/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 31 implementation notes, validation reports, and deployment documentation * Review of `package.json`, `demo-website/README.md`, `demo-website/public/demo/trend-finder-assets/README.md`, `docs/deployment.md`, `docs/ARCHITECTURE.md`, `docs/development.md`, and `README.md` * Scoped Prettier, markdownlint, ASCII/LF, and `git diff --check` validation for changed documentation files Remaining: * No Phase 31 repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` defines no unfinished phase after Phase 31, so `phasebuild` should not run unless a new phase is added to the PRD. ### 2026-06-22 - Phase 30 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 30 AI Rogue Game Extension release. * Verified Phase 30 Session 10 recorded first-release gate approval for AI Rogue. Phase 34 later superseded the visibility posture with production default enablement. * Updated root, architecture, extension, AI Rogue, and development docs to describe the shipped AI Rogue Play, Ledger, Loadout, Settings, PixiJS runtime, browser-local persistence, capped Insight Shards economy, and current non-goals. * Updated the extension package README so the registered extension list covers both Trend Finder and AI Rogue. * Replaced the stale Phase 29 spec-system documentation audit with the Phase 30 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `docs/README_docs.md` * `docs/development.md` * `docs/extensions/README_docs-extensions.md` * `docs/extensions/ai-rogue/README_ai-rogue.md` * `src/extensions/README_extensions.md` * `.spec_system/docs-audit.md` * `docs/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of Phase 30 implementation notes, validation reports, and enablement decision * Review of `package.json`, `src/extensions/ai-rogue/`, `src/assets/ai-rogue/`, `docs/extensions/ai-rogue/`, root docs, architecture docs, environment docs, development docs, and extension docs Remaining: * No Phase 30 repository documentation gaps remain from this pass. `.spec_system/PRD/PRD.md` defines no unfinished phase after Phase 30, so `phasebuild` should not run unless a new phase is added to the PRD. ### 2026-06-21 - Phase 29 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 29 TrendingAI comparison adoption release. * Verified the Trend Finder manuals already document the shipped Phase 29 projections, Reference mode checks, podcast/audio defer decision, broader social non-goal, and release validation evidence. * Updated the root README and architecture overview with the current comparison-derived Trend Finder surfaces. * Corrected the master PRD handoff from the stale post-Session 18 `audit` pointer to the post-documents final verification state. * Replaced the stale Phase 28 spec-system documentation audit with the Phase 29 audit report. Changed files: * `README.md` * `docs/ARCHITECTURE.md` * `.spec_system/PRD/PRD.md` * `.spec_system/docs-audit.md` * `docs/docs-audit.md` Validation: * `bash .spec_system/scripts/analyze-project.sh --json` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of `.spec_system/PRD/phase_29/PRD_phase_29.md` * Review of Phase 29 implementation notes and Session 17 defer stub * Review of root, architecture, development, extension, source, runtime, scoring, UI, security, and changelog docs Remaining: * Manual testing and an LLM audit remain recommended as final verification. `.spec_system/PRD/PRD.md` defines no unfinished phase after Phase 29, so `phasebuild` should not run unless a new phase is added to the PRD. ### 2026-06-09 - Phase 26 Documentation Closeout Check Completed: * Re-ran the documentation workflow for the completed Phase 26 Knowledge Graph Shared Brain release. * Verified the current source-of-truth docs already match the shipped release state: AI OS identity, Knowledge Graph surface, local API boundaries, architecture summary, root README version, and changelog entry were all current. * Left the primary documentation files unchanged because no stale claims or missing required docs were found during this pass. Changed files: * `docs/docs-audit.md` Validation: * `if [ -d ".spec_system/scripts" ]; then bash .spec_system/scripts/analyze-project.sh --json; else bash scripts/analyze-project.sh --json; fi` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of `README.md` * Review of `CONTRIBUTING.md` * Review of `docs/ARCHITECTURE.md` * Review of `docs/README_docs.md` * Review of `docs/project-identity.md` * Review of `docs/knowledge-graph/knowledge-graph.md` * Review of `docs/api/README_api.md` Remaining: * No follow-up documentation phase is required from this pass. The next workflow step is manual testing and LLM audit if the project owner wants an additional verification round. ### 2026-06-02 - Phase 22 Documents Pass Completed: * Re-checked the active Phase 22 documentation surface against the current PRD and verified the root README, architecture, onboarding, development, and environments docs remain current for the in-progress Antigravity and home-surface work. * Kept the source-of-truth docs unchanged because they already describe the implemented host / extension boundary accurately and the current session 03 scope is still planned rather than complete. Changed files: * `docs/docs-audit.md` Validation: * `if [ -d ".spec_system/scripts" ]; then bash .spec_system/scripts/analyze-project.sh --json; else bash scripts/analyze-project.sh --json; fi` * Review of `.spec_system/state.json` * Review of `.spec_system/PRD/PRD.md` * Review of `README.md`, `docs/README_docs.md`, `docs/ARCHITECTURE.md`, `docs/project-identity.md`, `docs/agent-pages.md`, `docs/local-agent-contract.md`, and `docs/CHANGELOG.md` Remaining: * Phase 22 is still in progress, so the next workflow step after any required manual testing and LLM audit remains `phasebuild` only if the PRD is later updated to mark the phase complete. ### 2026-06-02 - Phase 22 Documentation Sync Completed: * Reviewed the current phase 22 documentation surface against the active PRD and verified the existing core docs already describe the AI OS host / Trend Finder extension split, the Antigravity phase direction, and the in-progress phase 23 closeout scope accurately. * Synced the root README version line to the current package release `0.1.252`. * Left the existing architecture, onboarding, development, and environments docs unchanged because they already matched the implemented behavior and did not need speculative phase 22 detail. Changed files: * `README.md` * `docs/CHANGELOG.md` * `docs/docs-audit.md` Validation: * `node -p "require('./package.json').version"` * `rg -n "0\\.1\\.249|0\\.1\\.252" README.md docs/CHANGELOG.md` * Review of `.spec_system/PRD/PRD.md`, `.spec_system/PRD/phase_23/PRD_phase_23.md`, `docs/ARCHITECTURE.md`, `docs/onboarding.md`, `docs/development.md`, and `docs/environments.md` Remaining: * Phase 22 is still in progress, so the next workflow step after any required manual testing and LLM audit is not phasebuild unless the PRD is later updated to mark the phase complete. ### 2026-06-02 - Phase 21 Documentation Closeout Completed: * Reviewed the completed Phase 21 session summaries, validation reports, and current root documentation surface. * Confirmed the existing source-of-truth docs already describe the AI OS usage accuracy, project history, saved-time defaults, and Claude OAuth usage contracts accurately. * Updated the root README version to match the package release and added the missing Phase 21 Session 03 closeout entry to the changelog. Changed files: * `README.md` * `docs/CHANGELOG.md` * `docs/docs-audit.md` Validation: * `node -p "require('./package.json').version"` * Review of Phase 21 session summaries and validation reports * Stale-claim search for the README version line and missing Phase 21 Session 03 changelog entry Remaining: * None for the Phase 21 documentation closeout. ### 2026-06-01 - Install Audit Consolidation Completed: * Consolidated the completed ongoing install audit into canonical setup, testing, deployment, command, security, architecture, hackathon, script, and changelog docs. * Removed the ongoing-project install audit note after preserving the reusable certification, privacy, setup failure, isolated `HOME`, Playwright browser cache, and Worker preview details in existing docs. * Re-ran the inventory command and updated the current document-like file count after removing the obsolete note. Changed files: * `docs/ARCHITECTURE.md` * `docs/CHANGELOG.md` * `docs/SECURITY-COMPLIANCE.md` * `docs/commands.md` * `docs/deployment.md` * `docs/docs-audit.md` * `docs/hackathon/hackathon-submission.md` * Deleted obsolete ongoing install audit note * `docs/testing.md` * `scripts/README_scripts.md` Validation: * Inventory command from this document * Stale-claim search for the removed install audit note, old setup failure semantics, stale version claims, stale certification counts, and direct unpinned Wrangler commands * `bunx prettier --check` on touched docs Remaining: * None for the install audit consolidation. ### 2026-06-02 - Phase 19 Documentation Closeout Completed: * Reviewed the phase 19 session summaries, the archived phase tracker, and the current root documentation surface to confirm the Hermes Mission Control and Documents Gallery write features are already represented in the PRD and architecture docs. * Updated the root README version to match the current package release. * Added a phase 19 closeout entry to the changelog so the documentation trail reflects the completed Hermes orchestration and documents work. Changed files: * `README.md` * `docs/CHANGELOG.md` * `docs/docs-audit.md` Validation: * `node -p "require('./package.json').version"` * `rg -n "0\\.1\\.236|0\\.1\\.244|Mission Control|Documents Gallery" README.md docs/CHANGELOG.md .spec_system/PRD/PRD.md .spec_system/archive/phases/phase_19/PRD_phase_19.md` * Review of phase 19 implementation summaries and the archived phase PRD Remaining: * None for the phase 19 documentation closeout. ### 2026-06-02 - Phase 20 Documentation Closeout Completed: * Reviewed the completed Phase 20 Hermes long-tail and parity sign-off work against the current root README, architecture notes, agent page docs, API notes, environment matrix, and changelog. * Confirmed the Hermes Connections, derived stats, role map, CLI/terminal guidance, v2.3 route shell, public and token-gated bridge reads, admin gate boundaries, and demo-mode behavior are already documented in the source-of-truth docs. * No user-facing doc content changes were required because the existing docs were already current for the phase 20 closeout. Changed files: * `docs/docs-audit.md` Validation: * `rg -n "Hermes|connections|Overview|v2\\.3|admin gate|demo mode" README.md docs/ARCHITECTURE.md docs/agent-pages.md docs/api/README_api.md docs/environments.md docs/CHANGELOG.md .spec_system/PRD/PRD.md` * Review of `README.md`, `docs/ARCHITECTURE.md`, `docs/agent-pages.md`, `docs/api/README_api.md`, `docs/environments.md`, and `docs/CHANGELOG.md` against the phase 20 implementation notes Remaining: * Phase 21 is not started in the PRD, so the next workflow step after manual testing and LLM audit would be `phasebuild` if the phase transition continues. ### 2026-05-26 - Group 0 Audit Control Plane Completed: * Re-ran the inventory command and confirmed 87 document-like files outside ignored dependency/build directories. * Confirmed root symlink aliases remain `CLAUDE.md -> AGENTS.md` and `GEMINI.md -> AGENTS.md`. * Verified the group inventory against this file and found one unowned document: `docs/CODEOWNERS`. * Assigned `docs/CODEOWNERS` to Group 1 because it is a root project contract and repository ownership file. * Updated `docs/ongoing-projects/TODO.md` to remove the stale `X-line` placeholder for `scripts/aggregate.ts`. Changed files: * `docs/docs-audit.md` * `docs/ongoing-projects/TODO.md` * `docs/CHANGELOG.md` Validation: * `git status --short` * Inventory command from this document * Root symlink check with `find . -maxdepth 1 -type l` * `wc -l scripts/aggregate.ts` * `bun run format:check` Remaining: * Continue with Group 2 next per the suggested pass order, unless a future session intentionally chooses a different group. ### 2026-05-26 - Group 2 Product Source Of Truth Completed: * Audited `docs/PRD.md`, `docs/PRD_UX.md`, `docs/project-identity.md`, `docs/ARCHITECTURE.md`, `docs/CONVENTIONS.md`, `docs/CONSIDERATIONS.md`, and `docs/SECURITY-COMPLIANCE.md` against the current route, extension, script, env, and runtime boundaries. * Kept `docs/project-identity.md` unchanged because its AI OS host / Trend Finder extension naming rules still match the repository. * Updated `docs/PRD_UX.md` to use AI OS language for current host surfaces, document the implemented mobile sidebar sheet, replace stale OpenClaw concept preview language, and include the Engine Replay utility route. * Updated `docs/PRD.md` and `docs/ARCHITECTURE.md` so the source-of-truth route inventory includes `/extensions/trend-finder/engine` as a Trend Finder-scoped utility route rather than a creator-facing extension tab. * Updated `docs/ARCHITECTURE.md` with current local Vite middleware endpoints for manual Dream runs, saved-time config, and Trend Finder visibility config. * Updated `docs/CONSIDERATIONS.md` from Vite 7 to Vite 8 and `docs/CONVENTIONS.md` so test location guidance matches the current route/component/hook/script/e2e test layout. * Updated `docs/SECURITY-COMPLIANCE.md` env inventory for current Codex, Dream, Supabase, account-auth root, operator photo, and logging env names. Changed files: * `docs/PRD.md` * `docs/PRD_UX.md` * `docs/ARCHITECTURE.md` * `docs/CONVENTIONS.md` * `docs/CONSIDERATIONS.md` * `docs/SECURITY-COMPLIANCE.md` * `docs/docs-audit.md` Validation: * Source checks: `package.json`, `.env.local.example`, `src/routes/`, `src/components/app-sidebar.tsx`, `src/components/mobile-nav-trigger.tsx`, `src/components/ui/sidebar.tsx`, `src/extensions/trend-finder/client.tsx`, `src/routes/extensions.trend-finder.engine.tsx`, `vite.config.ts`, `scripts/lib/trend-finder-lens-bridge.ts`, `scripts/lib/trend-finder-visibility-config-bridge.ts`, and `src/server.ts`. * Stale-claim search for Vite 7, Claude OS current-host language, OpenClaw concept-preview language, hidden mobile-sidebar claims, and missing Engine Replay route coverage. * `bun run format:check` * `bun run typecheck` * `bun run test` Remaining: * Continue with Group 1 next per the suggested pass order so root docs can be synced from the refreshed source-of-truth set. ### 2026-05-26 - Group 1 Root Project Contract Completed: * Audited `README.md`, `AGENTS.md`, `CLAUDE.md`, `GEMINI.md`, `CONTRIBUTING.md`, `SECURITY.md`, `LICENSE`, and `docs/CODEOWNERS` against the refreshed source-of-truth docs and root setup/security sources. * Confirmed `CLAUDE.md` and `GEMINI.md` still point to `AGENTS.md`. * Kept `AGENTS.md`, `SECURITY.md`, and `LICENSE` unchanged because their identity, privacy, validation, and license boundaries still match the current project state. * Updated `README.md` so the useful script table includes current package scripts for development builds, Trend Finder backtests, backhistory replay, deferred Agents Codex smoke status, Dream cron compatibility installs, and the documented Vite preview caveat. * Updated `CONTRIBUTING.md` to describe `main` as the current private-project integration branch instead of promising production-ready code. * Updated `docs/CODEOWNERS` to remove stale hackathon-only ownership wording. Changed files: * `README.md` * `CONTRIBUTING.md` * `docs/CODEOWNERS` * `docs/docs-audit.md` Validation: * Verification sources: `package.json`, `.gitignore`, `.githooks/pre-commit`, `scripts/setup.ts`, `scripts/postinstall.sh`, `docs/project-identity.md`, `docs/CONVENTIONS.md`, and `docs/SECURITY-COMPLIANCE.md`. * Root symlink check with `find . -maxdepth 1 -type l`. * Hook state check with `git config --local --get core.hooksPath` and `.githooks/pre-commit` executable check. * Stale-claim search for Claude OS current-host language, OpenClaw concept preview language, Vite 7, setup/dev command drift, and root security/process overclaims. * ASCII check on touched files. * `bun run format:check` Remaining: * Continue with Group 3 next per the suggested pass order to audit setup, development, environment, deployment, CI/security, and testing docs. ### 2026-05-26 - Group 3 Setup, Development, Environments, Deployment Completed: * Audited `docs/onboarding.md`, `docs/development.md`, `docs/environments.md`, `docs/deployment.md`, `docs/deployment-blockers.md`, `docs/ci-security.md`, and `docs/testing.md` against package scripts, environment templates, Vite, Wrangler, Vitest, Playwright, GitHub Actions, gitleaks, and local hook configuration. * Updated onboarding prerequisites to name the Bun version floor and kept optional Trend Finder enablement language scoped to the extension. * Updated the development guide so its script table includes current package scripts for backhistory replay, account auth, Codex and Agents Codex smokes, Apify smoke checks, and hook reactivation. * Expanded environment documentation for current private source credentials, host AI runtime, Dream activation, OpenClaw path overrides, and the server-side operator photo source URL. Clarified that public `VITE_CLAUDE_OS_DREAM_ENABLED` is a setup preference, not scheduler activation. * Updated deployment docs and blockers so Wrangler local preview and manual deploy commands use the generated Worker config, and so CI deployment is documented as skipped when Cloudflare secrets are missing. * Updated CI security docs to use AI OS host language, include the current local hook provider prefixes and account-auth token checks, and document the local hook skip scope. * Updated testing docs to match the current Vitest include globs, Playwright web-server wrapper and env overrides, coverage exclusions, and enforced coverage thresholds. Changed files: * `docs/onboarding.md` * `docs/development.md` * `docs/environments.md` * `docs/deployment.md` * `docs/deployment-blockers.md` * `docs/ci-security.md` * `docs/testing.md` * `docs/docs-audit.md` Validation: * Verification sources: `package.json`, `.env.local.example`, `vite.config.ts`, `wrangler.jsonc`, `vitest.config.ts`, `playwright.config.ts`, `scripts/playwright-webserver.sh`, `.github/workflows/quality.yml`, `.github/workflows/integration.yml`, `.github/workflows/deploy.yml`, `.github/workflows/release.yml`, `.githooks/pre-commit`, and `.gitleaks.toml`. * Stale-claim search for outdated Trend Finder CI-security host language, missing package scripts, direct Wrangler preview/deploy commands, stale coverage thresholds, and missing test glob coverage. * ASCII check on touched Group 3 docs. * `bun run format:check` Remaining: * Continue with Group 4 next per the suggested pass order to audit host AI runtime, command, data-contract, agent-page, local-agent, skill-system, Apify, media-policy, and route-coverage docs. ### 2026-05-26 - Group 4 AI Runtime, Commands, Data, Agents, Skills Completed: * Audited `docs/ai-runtime-setup.md`, `docs/commands.md`, `docs/data-contract.md`, `docs/agent-pages.md`, `docs/local-agent-contract.md`, `docs/skill-system.md`, `docs/apify.md`, `docs/media-policy.md`, and `docs/route-coverage.md` against current package scripts, runtime config, route tree, live-data contract, local-agent assembly, OpenClaw bridge, skill scanner/setup behavior, Apify integration, assets, and tests. * Updated AI runtime docs to distinguish code defaults from example host and Trend Finder compatibility overrides. * Expanded command docs so all 33 `package.json` scripts are documented, including postinstall, backhistory replay, OpenAI account auth, Codex smoke, deferred Agents Codex smoke, Apify smoke, and Playwright e2e coverage. * Updated data-contract docs for current `openclaw`, `localAgents`, and `usage.codexWindow` branches, and clarified that provider-neutral activity uses `localAgents.sessions`. * Updated agent and local-agent docs for the implemented OpenClaw header status pill, current OpenClaw data/status states, and migrated-route guidance. * Updated skill docs for `CODEX_HOME` support and the two bundled local skill assets, Dream and personas. * Updated Apify, media-policy, and route-coverage docs for current validation commands, public served assets, 15 generated route entries, Engine Replay coverage, 204 Vitest files, and 8 Playwright specs. Changed files: * `docs/ai-runtime-setup.md` * `docs/commands.md` * `docs/data-contract.md` * `docs/agent-pages.md` * `docs/local-agent-contract.md` * `docs/skill-system.md` * `docs/apify.md` * `docs/media-policy.md` * `docs/route-coverage.md` * `docs/docs-audit.md` Validation: * Verification sources: `package.json`, `.env.local.example`, `scripts/postinstall.sh`, `scripts/trend-finder-replay-backhistory.ts`, `scripts/openai-account-auth.mjs`, `scripts/openai-codex-smoke.mjs`, `scripts/openai-agents-codex-smoke.mjs`, `scripts/apify-smoke.ts`, `scripts/lib/`, `src/lib/`, `src/routes/`, `playwright.config.ts`, `scripts/playwright-webserver.sh`, and `skills/`. * Package-script coverage check confirmed 33 documented command sections for 33 package scripts, with no missing or extra entries. * Route-tree source check confirmed `FileRoutesByFullPath` has 15 entries and includes `/extensions/trend-finder/engine`. * Test inventory check confirmed 204 Vitest files, with 105 under `src` and 99 under `scripts`, plus 8 Playwright specs under `tests/e2e/`. * Stale-claim search for old route counts, old test counts, stale OpenClaw follow-up language, stale Phase 09 migration language, old Apify test commands, old Codex skill root wording, and old AI runtime default headings. * ASCII check on touched Group 4 docs. * `bun run test -- src/routes/__tests__/route-tree.test.ts src/routes/__tests__/extensions-routes.test.tsx src/routes/__tests__/root-component.test.tsx src/lib/__tests__/validate-live-data.test.ts src/lib/__tests__/validate-live-data-extensions.test.ts scripts/lib/__tests__/local-agent-assembly.test.ts scripts/lib/__tests__/openclaw-scanner.test.ts scripts/lib/__tests__/skill-scanner.test.ts scripts/lib/__tests__/ai-runtime-config.test.ts scripts/lib/apify/__tests__ scripts/extensions/trend-finder/sources/__tests__/apify-source-config.test.ts scripts/extensions/trend-finder/sources/__tests__/apify-smoke.test.ts` passed 16 files and 207 tests. * `bun run format:check` Remaining: * Continue with Group 5 next per the suggested pass order to audit the extension overview and Trend Finder concept docs. ### 2026-05-26 - Group 5 Extension Overview And Trend Finder Concepts Completed: * Audited `docs/extensions/README_docs-extensions.md`, `docs/extensions/trend-finder/concepts.md`, `docs/extensions/trend-finder/history.md`, `docs/extensions/trend-finder/pipeline.md`, `docs/extensions/trend-finder/runtime-and-provenance.md`, and `docs/extensions/trend-finder/ui-surfaces.md` against the current Trend Finder extension, route, collector, visibility, Engine Replay Reference, and browser tests. * Updated extension and concept docs so Engine Replay is described as a two-mode route: Replay for latest-run proof and Reference for in-app Markdown manuals. * Updated pipeline docs with current visibility state locations, the optional loopback file overlay, Engine Reference view, reference-doc registry, and visibility bridge implementation files. * Updated runtime/provenance docs so Replay stage descriptions include movement analysis, prediction writer, retro evaluator, and artifact writing, while Reference mode is explicitly documentation-only. * Updated UI surface docs with current per-view visibility controls, storage key, optional file config endpoint, and Engine Replay route mode boundaries. * Kept `docs/extensions/trend-finder/history.md` unchanged because its topic identity, movement, prediction/retro, and script-only backtest claims still match the current implementation. Changed files: * `docs/extensions/README_docs-extensions.md` * `docs/extensions/trend-finder/concepts.md` * `docs/extensions/trend-finder/pipeline.md` * `docs/extensions/trend-finder/runtime-and-provenance.md` * `docs/extensions/trend-finder/ui-surfaces.md` * `docs/docs-audit.md` Validation: * Verification sources: `src/extensions/trend-finder/client.tsx`, `src/extensions/trend-finder/schema.ts`, `src/extensions/trend-finder/view-model.ts`, `src/extensions/trend-finder/reference-docs.ts`, `src/extensions/trend-finder/visibility-config.ts`, `src/extensions/trend-finder/use-visibility-settings.ts`, `src/extensions/trend-finder/views/engine-reference-view.tsx`, `src/routes/extensions.trend-finder.engine.tsx`, `scripts/extensions/trend-finder/collector.ts`, `scripts/lib/trend-finder-visibility-config-bridge.ts`, and `vite.config.ts`. * Source checks confirmed Trend Finder still has five creator views, Engine Replay is a dedicated route with Replay and Reference modes, Reference mode renders eight committed Trend Finder Markdown manuals, and visibility settings use browser localStorage plus an optional loopback file overlay. * Inventory command from this document now reports 84 existing document-like files. The lower count reflects three currently deleted Group 9 planning records in the worktree; Group 9 should resolve their final disposition. * Stale-claim search for missing Engine Replay Reference mode, missing visibility storage/config docs, stale static Engine route wording, stale route verification path, and missing source-stage terminology. * ASCII check on touched Group 5 docs. * `bun run test -- src/extensions/trend-finder/__tests__/reference-docs.test.ts src/routes/__tests__/extensions-routes.test.tsx src/extensions/trend-finder/__tests__/visibility-config.test.ts src/extensions/trend-finder/__tests__/visibility-views.test.tsx` passed 4 files and 35 tests. * `bun run format:check` Remaining: * Continue with Group 6 next per the suggested pass order to audit Trend Finder scoring, source compliance, and Creator Lens docs. ### 2026-05-26 - Group 6 Trend Finder Scoring, Sources, Creator Lens Completed: * Audited `docs/extensions/trend-finder/scoring.md`, `docs/extensions/trend-finder/sources.md`, `docs/extensions/trend-finder/creator-lens.md`, and the seven `docs/sources/source-compliance-*.md` files against the current scoring, watchlist, source declaration, source adapter, normalizer, Creator Lens, snapshot, backtest, env, and cache boundaries. * Kept `docs/extensions/trend-finder/scoring.md` unchanged because the current score weights, hidden-gem thresholds, watchlist rules, source status weights, source role weights, low-quality cap, and fallback lens effects still match code. * Updated `docs/extensions/trend-finder/sources.md` with the current source-related env names and the implemented generated-data plus private snapshot deletion boundary. * Updated `docs/extensions/trend-finder/creator-lens.md` with the current browser draft storage key, legacy draft-key migration, and token-gated loopback bridge boundary. * Updated source compliance docs so current browser-visible source identifiers match configured source IDs, retention/deletion paths include private Trend Finder snapshots, HN no longer says it still awaits adapter enablement, and Product Hunt describes the implemented private backtest archive path without claiming browser-visible historical evidence output. Changed files: * `docs/extensions/trend-finder/sources.md` * `docs/extensions/trend-finder/creator-lens.md` * `docs/sources/source-compliance-arxiv.md` * `docs/sources/source-compliance-github.md` * `docs/sources/source-compliance-hackernews.md` * `docs/sources/source-compliance-producthunt.md` * `docs/sources/source-compliance-reddit.md` * `docs/sources/source-compliance-rss-news.md` * `docs/sources/source-compliance-youtube.md` * `docs/docs-audit.md` Validation: * Verification sources: `.env.local.example`, `.gitignore`, `scripts/lib/ai-runtime/scoring.ts`, `scripts/lib/ai-runtime/watchlist.ts`, `scripts/lib/ai-runtime/snapshots.ts`, `scripts/lib/ai-runtime/backtests.ts`, `scripts/extensions/trend-finder/sources/apify-source-config.ts`, `scripts/extensions/trend-finder/sources/source-compliance.ts`, `scripts/extensions/trend-finder/sources/source-quality.ts`, `scripts/extensions/trend-finder/sources/apify-adapter.ts`, `scripts/extensions/trend-finder/sources/apify-normalizers.ts`, `scripts/extensions/trend-finder/sources/hn-adapter.ts`, `scripts/extensions/trend-finder/sources/health.ts`, `scripts/lib/trend-finder-lens-bridge.ts`, `src/extensions/trend-finder/creator-lens-storage.ts`, and `docs/runbooks/trend-finder-backtests.md`. * Source checks confirmed one built-in HN source, six reviewed Apify source declarations, exact configured source IDs, current source env names, Product Hunt as the only reviewed historical source, and `.cache/` as the ignored private snapshot/backtest root. * Stale-claim search for old short source identifiers, live-data-only retention claims, missing snapshot deletion paths, obsolete HN enablement wording, and obsolete Product Hunt "no persistent backtest storage" wording. * ASCII check on touched Group 6 docs. * `bun run test -- scripts/lib/ai-runtime/__tests__/scoring.test.ts scripts/lib/ai-runtime/__tests__/watchlist.test.ts scripts/extensions/trend-finder/sources/__tests__/apify-source-config.test.ts scripts/extensions/trend-finder/sources/__tests__/apify-normalizers.test.ts scripts/extensions/trend-finder/sources/__tests__/hn-adapter.test.ts` passed 5 files and 62 tests. * `bunx prettier --check docs/extensions/trend-finder/scoring.md docs/extensions/trend-finder/sources.md docs/extensions/trend-finder/creator-lens.md docs/sources/source-compliance-arxiv.md docs/sources/source-compliance-github.md docs/sources/source-compliance-hackernews.md docs/sources/source-compliance-producthunt.md docs/sources/source-compliance-reddit.md docs/sources/source-compliance-rss-news.md docs/sources/source-compliance-youtube.md docs/docs-audit.md` Remaining: * Continue with Group 7 next per the suggested pass order to audit runbooks and operations docs. ### 2026-05-26 - Group 7 Runbooks And Operations Completed: * Audited `docs/runbooks/ai-os-dream.md`, `docs/runbooks/incident-response.md`, `docs/runbooks/scheduled-aggregate.md`, and `docs/runbooks/trend-finder-backtests.md` against the current scheduler, Dream, backtest, backhistory replay, package script, and gitignore boundaries. * Updated the Dream runbook so CLI execution distinguishes an already activated scheduler process from one-shot activation with `AI_OS_DREAM_ENABLED=true bun run scheduler:dream:run`. * Updated the Dream scheduling example so user-created timers include explicit Dream activation while preserving the disabled-by-default default. * Expanded incident response with scheduler blocked/stale-state recovery, Dream skipped/failed recovery, Trend Finder backtest and private backhistory replay recovery, and private artifact sharing boundaries. * Updated the Trend Finder backtest runbook to cover the implemented `trend-finder:replay-backhistory` command, private backhistory input expectations, private archive outputs, optional current-history seeding, and bounded aggregate publishing. * Kept `docs/runbooks/scheduled-aggregate.md` unchanged because its aggregate cadence, status commands, timeout, private scheduler paths, and Dream boundary still match the current scheduler implementation. Changed files: * `docs/runbooks/ai-os-dream.md` * `docs/runbooks/incident-response.md` * `docs/runbooks/trend-finder-backtests.md` * `docs/docs-audit.md` Validation: * Verification sources: `package.json`, `.gitignore`, `scripts/scheduler-runner.ts`, `scripts/scheduler-status.ts`, `scripts/install-dream-cron.ts`, `scripts/dashboard-dream-run.ts`, `src/lib/async-dream-run.ts`, `vite.config.ts`, `scripts/lib/scheduler/registry.ts`, `scripts/lib/scheduler/paths.ts`, `scripts/lib/scheduler/operator-status.ts`, `scripts/lib/scheduler/aggregate-handler.ts`, `scripts/lib/scheduler/dream-handler.ts`, `scripts/trend-finder-backtest.ts`, `scripts/trend-finder-replay-backhistory.ts`, and `scripts/lib/ai-runtime/backtests.ts`. * Command help checks: `bun run scripts/scheduler-status.ts --help`, `bun run trend-finder:backtest -- --help`, and `bun run trend-finder:replay-backhistory -- --help`. * Stale-claim checks for Dream manual activation, scheduler status commands, scheduler private paths, backtest source support, and missing backhistory replay runbook coverage. * ASCII check on touched Group 7 docs. * `bun run test -- scripts/lib/__tests__/scheduler-registry.test.ts scripts/lib/__tests__/scheduler-status-cli.test.ts scripts/lib/__tests__/scheduler-runner.test.ts scripts/lib/__tests__/scheduler-dream-handler.test.ts scripts/lib/ai-runtime/__tests__/backtests.test.ts scripts/extensions/trend-finder/__tests__/historical-backhistory-flow.test.ts` passed 6 files and 57 tests. * `bun run format:check` Remaining: * Continue with Group 8 next per the suggested pass order to audit hackathon, public demo, and research docs. ### 2026-05-26 - Group 8 Hackathon, Public Demo, And Research Notes Completed: * Audited `docs/hackathon/brainstorm-hackathon.md`, `docs/hackathon/hackathon-project-brief.md`, `docs/hackathon/hackathon-submission.md`, `docs/hackathon/trend-finder-demo.md`, and `docs/research/parallel-projects.md` against the current Trend Finder dashboard, run control, Creator Lens, watchlist, movement, source context, evidence metric, Engine Replay, and demo boundaries. * Updated the brainstorm record so its old Phase 05 gaps are clearly historical and its current reconciliation reflects editable Creator Lens saves, local Run Trend Finder control, generated watchlist rows, per-topic source context, evidence metric chips, movement/previous-score display, predictions, retros, and script-only historical backtests. * Updated the hackathon brief, submission guide, and demo workflow so Loom proof points include the local run control, generated watchlist, source context, public metric chips, movement labels, and Engine Replay prediction/retro stages without calling them hosted, unrestricted, or live by default. * Changed the parallel-projects research doc from a "current external research snapshot" claim to a point-in-time snapshot that was not externally refreshed during this audit. Changed files: * `docs/hackathon/brainstorm-hackathon.md` * `docs/hackathon/hackathon-project-brief.md` * `docs/hackathon/hackathon-submission.md` * `docs/hackathon/trend-finder-demo.md` * `docs/research/parallel-projects.md` * `docs/docs-audit.md` Validation: * Verification sources: `README.md`, `docs/project-identity.md`, `src/extensions/trend-finder/client.tsx`, `src/extensions/trend-finder/components/creator-lens-panel.tsx`, `src/extensions/trend-finder/hooks/use-creator-lens-draft.ts`, `src/extensions/trend-finder/hooks/use-trend-finder-run.ts`, `src/extensions/trend-finder/components/trend-card.tsx`, `src/extensions/trend-finder/components/evidence-links.tsx`, `src/extensions/trend-finder/components/source-breakdown.tsx`, `src/extensions/trend-finder/views/watchlist-view.tsx`, `src/routes/extensions.trend-finder.engine.tsx`, `scripts/extensions/trend-finder/collector.ts`, `scripts/lib/ai-runtime/watchlist.ts`, `scripts/lib/trend-finder-lens-bridge.ts`, and `vite.config.ts`. * Stale-claim search for read-only Creator Lens language, missing run-control language, empty watchlist output, missing source breakdowns, missing evidence metrics, and missing previous-score/movement display. * ASCII check on touched Group 8 docs and this work file. * Inventory command from this document still reports 84 document-like files. * `bun run test -- src/lib/__tests__/trend-finder-dashboard.test.tsx src/lib/__tests__/trend-finder-schema.test.ts src/extensions/trend-finder/__tests__/view-model.test.ts scripts/extensions/trend-finder/__tests__/collector.test.ts scripts/lib/ai-runtime/__tests__/watchlist.test.ts` passed 5 files and 133 tests. * `bun run format:check` Remaining: * Continue with Group 9 next per the suggested pass order to audit ADRs, API notes, and ongoing planning records, including the three currently deleted Group 9 planning docs in the worktree. ### 2026-05-26 - Group 9 ADRs, API, And Planning Records Completed: * Audited `docs/adr/0000-template.md`, `docs/adr/0001-extension-platform-foundation.md`, `docs/api/README_api.md`, and `docs/ongoing-projects/sqlite-observation-store-transition-plan.md` against current endpoint, extension, source, cache, route, and planning-record boundaries. * Added a maintenance note to ADR 0001 so its Phase 02 source-status wording is clearly historical and current Trend Finder source/runtime docs remain the owner for implemented behavior. * Updated the Local API Notes Trend Finder API boundary so it includes both implemented dev-only Trend Finder HTTP config surfaces: `/__trend-finder-visibility-config` and `/__trend_finder_creator_lens`. * Kept the SQLite observation store transition plan as planned work because no AI OS observation-store implementation exists yet; current SQLite usage is limited to read-only OpenClaw task-run scanning. * Verified the three deleted Group 9 planning records have no references outside this work file and treated them as retired implemented-plan records rather than restoring them to active docs maintenance. Changed files: * `docs/adr/0001-extension-platform-foundation.md` * `docs/api/README_api.md` * `docs/docs-audit.md` Retired records already deleted in the worktree: * `docs/ongoing-projects/engine-replay-reference-tabs-plan.md` * `docs/ongoing-projects/skill-saved-time-file-overrides-plan.md` * `docs/ongoing-projects/trend-finder-card-visibility-plan.md` Validation: * Inventory command from this document still reports 84 document-like files. * Root symlink check with `find . -maxdepth 1 -type l`. * Endpoint/source checks against `vite.config.ts`, `src/server.ts`, `scripts/lib/trend-finder-visibility-config-bridge.ts`, `scripts/lib/trend-finder-lens-bridge.ts`, `scripts/lib/hermes-dev-bridge.ts`, `scripts/lib/hermes-admin-bridge.ts`, `scripts/lib/openclaw-dev-bridge.ts`, and `scripts/lib/openclaw-admin-bridge.ts`. * Reference search for deleted Group 9 planning records and active Group 9 docs. * Stale-claim search for Trend Finder API boundary wording and SQLite observation-store implementation claims. * ASCII check on touched Group 9 docs and this work file. * `bun run format:check` Remaining: * Continue with Group 10 next per the suggested pass order to audit directory READMEs. ### 2026-05-26 - Group 10 Directory READMEs Completed: * Audited `data/README_data.md`, `logs/README_logs.md`, `public/README_public.md`, `public/assets/trend-finder/README.md`, `scripts/README_scripts.md`, `scripts/extensions/README_extensions-scripts.md`, `skills/README_skills.md`, `src/README_src.md`, `src/extensions/README_extensions.md`, and `tests/README_tests.md` against current directory contents, package scripts, `.gitignore`, setup behavior, public asset usage, extension routes, and test layout. * Kept `data/README_data.md`, `public/README_public.md`, `scripts/extensions/README_extensions-scripts.md`, `src/README_src.md`, and `tests/README_tests.md` unchanged because their ownership, privacy, and validation boundaries still match the repository. * Updated `logs/README_logs.md` and `scripts/README_scripts.md` so generated logs are documented as ignored by the root `.gitignore` with only `logs/README_logs.md` tracked. * Updated the Trend Finder public asset README with a proper directory guide, current image inventory, active hero/background wording, root-relative URL mapping, and private-artifact boundary. * Updated `skills/README_skills.md` to include both bundled local skills, setup destinations, Hermes no-clobber behavior, and the current `CODEX_HOME` non-install boundary. * Updated `src/extensions/README_extensions.md` so the extension platform is described as AI OS host infrastructure, with Trend Finder as the current registered extension and Engine Replay as a dedicated utility route rather than a creator-facing extension view. * Updated `scripts/README_scripts.md` to include current script files and scheduler, Dream, backtest, postinstall, Playwright, asset, bundle, and transport helper surfaces while linking `docs/commands.md` as the owner for the full package script catalog. Changed files: * `logs/README_logs.md` * `public/assets/trend-finder/README.md` * `scripts/README_scripts.md` * `skills/README_skills.md` * `src/extensions/README_extensions.md` * `docs/docs-audit.md` Validation: * Inventory command from this document still reports 84 document-like files. * Directory checks against `data/`, `logs/`, `public/`, `public/assets/trend-finder/`, `scripts/`, `scripts/extensions/`, `skills/`, `src/`, `src/extensions/`, and `tests/`. * Verification sources: `package.json`, `.gitignore`, `scripts/setup.ts`, `scripts/lib/skill-scanner.ts`, `docs/commands.md`, `docs/skill-system.md`, `docs/testing.md`, `docs/media-policy.md`, `src/extensions/trend-finder/client.tsx`, and `src/routes/extensions.trend-finder.engine.tsx`. * Stale-claim search for missing log sentinel wording, old submission path, outdated hero wording, historical phase-only certification wording, and Trend Finder-specific extension-platform wording. * ASCII check on touched Group 10 docs and this work file. * `bunx prettier --write logs/README_logs.md public/assets/trend-finder/README.md scripts/README_scripts.md skills/README_skills.md src/extensions/README_extensions.md` * `bun run format:check` Remaining: * Continue with Group 11 next per the suggested pass order to audit local skill docs. ### 2026-05-26 - Group 11 Local Skill Docs Completed: * Audited `skills/dream/SKILL.md` and `skills/personas/SKILL.md` against setup/install behavior, the current AI OS Dream scheduler, Dream output validation/loading, and local Hermes persona boundaries. * Updated `skills/dream/SKILL.md` so the inherited `/dream` skill is clearly a legacy compatibility asset, while the current AI OS Dream scheduler remains the primary control plane writing to `~/.ai-os/dreams/`. * Updated the legacy Dream output instructions from the old `cat`/`headline`/`prescription`-centered shape to the current AI OS Dream validator contract requiring `id`, `title`, and `summary`, with optional browser-safe legacy display hints. * Updated Dream metadata guidance to use only the current allowlisted metadata keys so legacy fallback output can be accepted by the AI OS loader. * Kept `skills/personas/SKILL.md` unchanged because its read-only `~/.hermes/pantheon/personas/` resolution and secret-handling boundaries still match the bundled local skill purpose. Changed files: * `skills/dream/SKILL.md` * `docs/docs-audit.md` Validation: * Inventory command from this document still reports 84 document-like files. * Verification sources: `scripts/setup.ts`, `scripts/install-dream-cron.ts`, `scripts/lib/dream/paths.ts`, `scripts/lib/dream/output-validation.ts`, `scripts/lib/dream/loader.ts`, `scripts/lib/dream/execution.ts`, `src/lib/home-transforms.ts`, `src/data/live-data.example.json`, `skills/README_skills.md`, `docs/skill-system.md`, and `docs/runbooks/ai-os-dream.md`. * Stale-claim search for legacy Dream output shape, legacy Dream state paths, current AI OS Dream scheduler ownership, and local skill install paths. * ASCII check on touched Group 11 docs and this work file. * `bun run test -- scripts/lib/__tests__/dream-writes.test.ts scripts/lib/__tests__/dream-loader.test.ts scripts/lib/__tests__/dream-request-builder.test.ts scripts/lib/__tests__/scheduler-dream-handler.test.ts` passed 4 files and 17 tests. * `bunx prettier --write skills/dream/SKILL.md docs/docs-audit.md` * `bun run format:check` * `git diff --check -- docs/docs-audit.md skills/dream/SKILL.md` Remaining: * Continue with Group 12 next per the suggested pass order to disposition imported Claude OS example docs. ### 2026-05-26 - Group 12 Example And Imported Reference Docs Completed: * Audited the ignored imported Claude OS example bundle under `EXAMPLES/claudeos/`, including the v1 and v2 README, CLAUDE, LICENSE, Dream skill, personas skill, and diff files listed in the Group 12 inventory. * Confirmed `EXAMPLES/` is ignored by `.gitignore`, has no tracked files, and is excluded from ESLint, Vitest, Playwright, Vite watcher, and coverage scopes. * Confirmed active tracked docs do not depend on internal `EXAMPLES/` paths or imported example instructions; current tracked references describe only the imported Claude OS foundation, external Original Claude OS source, and license/ownership boundary. * Treated the imported bundle as an archival/reference snapshot, not active AI OS documentation. No edits were made inside `EXAMPLES/` so upstream example content and license texts remain intact. Changed files: * `docs/docs-audit.md` Validation: * Inventory command from this document still reports 84 document-like files. * Verification sources: `.gitignore`, `eslint.config.js`, `vitest.config.ts`, `playwright.config.ts`, `vite.config.ts`, `README.md`, `LICENSE`, `docs/project-identity.md`, and `docs/PRD.md`. * Import reference search excluding `EXAMPLES/**` confirmed no active docs link to internal imported example paths. * Imported bundle check confirmed `EXAMPLES/claudeos/DIFF_V1_V2_summary.json` exists alongside `DIFF_V1_V2.md`. Remaining: * Continue with Group 13 next per the suggested pass order to disposition the generated dogfood audit report. ### 2026-05-26 - Group 13 Generated Or Ignored Audit Artifacts Completed: * Audited the generated dogfood bundle under `dogfood-output/trend-finder-20260525-090500/` as an ignored evidence snapshot, not active AI OS or Trend Finder product documentation. * Confirmed the bundle contains `report.md`, screenshots, text captures, and videos from the 2026-05-25 Trend Finder dogfood pass, and left the generated contents unchanged. * Confirmed `dogfood-output/` is ignored by `.gitignore`, has no tracked files, and is covered by the private runtime artifact guardrail check. * Confirmed current tracked docs do not depend on generated report paths, issue titles, or report claims outside this audit tracker. * Verified `docs/hackathon/trend-finder-demo.md`, `docs/testing.md`, and `tests/README_tests.md` already keep generated browser artifacts and demo report language scoped correctly. Changed files: * `docs/docs-audit.md` Validation: * Inventory command from this document still reports 84 document-like files. * Generated artifact inventory with `find dogfood-output -maxdepth 4 -type f`. * Ignore checks with `git check-ignore -v` for `dogfood-output/`, `.cache/`, and `EXAMPLES/` paths. * Tracking check with `git ls-files dogfood-output .cache EXAMPLES`. * Reference search excluding generated and imported artifact directories confirmed no active tracked docs depend on the generated dogfood report. * `bun run runtime:check-private` Remaining: * Final Group 0 wrap is complete. No documentation audit groups remain open. ### 2026-05-26 - Final Group 0 Audit Wrap Completed: * Re-ran the documented inventory command and confirmed the current snapshot is still 84 document-like files outside ignored dependency/build directories. * Confirmed every inventory path is represented in this audit tracker. * Confirmed root symlink aliases remain `CLAUDE.md -> AGENTS.md` and `GEMINI.md -> AGENTS.md`. * Confirmed all audit groups, including imported examples and generated audit artifacts, are checked off and have recorded dispositions. * Updated `docs/CHANGELOG.md` with the final documentation audit outcome. Changed files: * `docs/docs-audit.md` * `docs/CHANGELOG.md` Validation: * Inventory command from this document. * Root symlink check with `find . -maxdepth 1 -type l`. * Inventory coverage check comparing each current inventory path against this audit tracker. * ASCII check on `docs/docs-audit.md` and `docs/CHANGELOG.md`. * `git diff --check -- docs/docs-audit.md docs/CHANGELOG.md` * `bun run format:check` Remaining: * None. Future documentation changes should use this file as the next audit control plane and start by refreshing the inventory count. ## Per-Group Audit Checklist For each group: * Read the docs in the group and the smallest useful set of source files they claim to describe. * Check package scripts, route files, script files, schemas, env examples, and tests only when the doc makes claims about them. * Mark every claim as current, stale, duplicated, planned, historical, or unverifiable. * Update stale current-claim language immediately. * Convert broad future claims into explicit deferred/planned language. * Link to the source-of-truth doc when content is duplicated. * Do not broaden scope into another group unless the current doc cannot be made accurate without that handoff. ## Audit Groups ### Group 0 - Audit Control Plane Goal: Keep the audit process itself current before touching product docs. Files: * `docs/docs-audit.md` * `docs/CHANGELOG.md` * `docs/ongoing-projects/TODO.md` Verification sources: * `git status --short` * `package.json` * Latest inventory command output Exit criteria: * The inventory count is current. * Audit groups still cover every document-like file. * New audit findings have an owner group. ### Group 1 - Root Project Contract Goal: Verify the top-level project identity, quickstart, security posture, and agent handoff docs. Files: * `README.md` * `AGENTS.md` * `CLAUDE.md` symlink alias * `GEMINI.md` symlink alias * `CONTRIBUTING.md` * `SECURITY.md` * `LICENSE` * `docs/CODEOWNERS` Verification sources: * `package.json` * `.gitignore` * `.githooks/` * `scripts/setup.ts` * `scripts/postinstall.sh` Exit criteria: * One-command local start remains accurate. * AI OS and Trend Finder naming is aligned with `docs/project-identity.md`. * Security and contribution instructions do not promise unavailable processes. ### Group 2 - Product Source Of Truth Goal: Audit high-level product, UX, architecture, identity, and convention docs as the upstream reference set for the rest of the documentation. Files: * `docs/PRD.md` * `docs/PRD_UX.md` * `docs/project-identity.md` * `docs/ARCHITECTURE.md` * `docs/CONVENTIONS.md` * `docs/CONSIDERATIONS.md` * `docs/SECURITY-COMPLIANCE.md` Verification sources: * `src/routes/` * `src/extensions/` * `src/lib/` * `scripts/` * `docs/CHANGELOG.md` Exit criteria: * Current phase status, implemented behavior, and deferred work are separated. * Architecture docs match the actual route, script, and extension boundaries. * Privacy/security posture matches current code behavior. ### Group 3 - Setup, Development, Environments, Deployment Goal: Verify operator setup, local development, environment variables, deployment state, blockers, CI/security, and testing instructions. Files: * `docs/onboarding.md` * `docs/development.md` * `docs/environments.md` * `docs/deployment.md` * `docs/deployment-blockers.md` * `docs/ci-security.md` * `docs/testing.md` Verification sources: * `package.json` * `.github/workflows/` * `.env.local.example` * `wrangler.jsonc` * `vite.config.ts` * `playwright.config.ts` * `vitest.config.ts` Exit criteria: * Setup and validation commands use Bun and match package scripts. * Env docs distinguish browser-safe values from secrets. * Deployment docs do not imply production readiness beyond implemented support. ### Group 4 - AI Runtime, Commands, Data, Agents, Skills Goal: Audit host-platform runtime docs and command docs without mixing them with Trend Finder extension claims. Files: * `docs/ai-runtime-setup.md` * `docs/commands.md` * `docs/data-contract.md` * `docs/agent-pages.md` * `docs/local-agent-contract.md` * `docs/skill-system.md` * `docs/apify.md` * `docs/media-policy.md` * `docs/route-coverage.md` Verification sources: * `package.json` * `scripts/` * `scripts/lib/` * `src/lib/` * `src/routes/` * `skills/` Exit criteria: * Commands, env names, and runtime boundaries match the current implementation. * AI OS host docs stay separate from Trend Finder source, scoring, and demo docs. * Route and data-contract docs match current schemas and route files. ### Group 5 - Extension Overview And Trend Finder Concepts Goal: Audit extension documentation as a coherent Trend Finder source of truth. Files: * `docs/extensions/README_docs-extensions.md` * `docs/extensions/trend-finder/concepts.md` * `docs/extensions/trend-finder/history.md` * `docs/extensions/trend-finder/pipeline.md` * `docs/extensions/trend-finder/runtime-and-provenance.md` * `docs/extensions/trend-finder/ui-surfaces.md` Verification sources: * `src/extensions/` * `src/routes/extensions.*.tsx` * `src/lib/trend-finder*` * `scripts/extensions/` * `scripts/lib/trend-finder-visibility-config-bridge.ts` * `src/data/live-data.example.json` Exit criteria: * Extension docs describe implemented extension contracts and runtime data only. * Trend Finder terminology is consistent across pipeline, UI, and provenance docs. * Historical behavior is clearly labeled as historical. ### Group 6 - Trend Finder Scoring, Sources, Creator Lens Goal: Audit the detailed Trend Finder mechanics and public source compliance docs together, because stale scoring/source claims can cascade quickly. Files: * `docs/extensions/trend-finder/scoring.md` * `docs/extensions/trend-finder/sources.md` * `docs/extensions/trend-finder/creator-lens.md` * `docs/sources/source-compliance-arxiv.md` * `docs/sources/source-compliance-github.md` * `docs/sources/source-compliance-hackernews.md` * `docs/sources/source-compliance-producthunt.md` * `docs/sources/source-compliance-reddit.md` * `docs/sources/source-compliance-rss-news.md` * `docs/sources/source-compliance-youtube.md` Verification sources: * `scripts/extensions/sources/` * `scripts/extensions/` * `src/extensions/trend-finder/` * `src/lib/trend-finder*` * Environment key names in `.env.local.example` Exit criteria: * Source status, rate-limit notes, retention notes, and env names are current. * Blocked, deferred, fixture-backed, and live-capable sources are labeled accurately. * Scoring docs match code or explicitly identify planned scoring behavior. ### Group 7 - Runbooks And Operations Goal: Verify operational procedures, private artifact boundaries, scheduler behavior, and incident response. Files: * `docs/runbooks/ai-os-dream.md` * `docs/runbooks/incident-response.md` * `docs/runbooks/scheduled-aggregate.md` * `docs/runbooks/trend-finder-backtests.md` Verification sources: * `scripts/scheduler-runner.ts` * `scripts/scheduler-status.ts` * `scripts/install-dream-cron.ts` * `scripts/trend-finder-backtest.ts` * `scripts/trend-finder-replay-backhistory.ts` * `.gitignore` Exit criteria: * Runbook commands match package scripts. * Private artifacts are described by path/policy, not by inspecting private data. * Failure modes and recovery steps match current scripts. ### Group 8 - Hackathon, Public Demo, And Research Notes Goal: Keep public/demo-facing Trend Finder docs honest about what is current, historical, brainstormed, or submission-specific. Files: * `docs/hackathon/brainstorm-hackathon.md` * `docs/hackathon/hackathon-project-brief.md` * `docs/hackathon/hackathon-submission.md` * `docs/hackathon/trend-finder-demo.md` * `docs/research/parallel-projects.md` Verification sources: * `README.md` * `docs/project-identity.md` * `src/extensions/trend-finder/` * `src/routes/extensions.$extensionId.tsx` * `src/routes/extensions.$extensionId.$viewId.tsx` * `src/routes/extensions.trend-finder.engine.tsx` Exit criteria: * Public/demo docs do not overclaim live ingestion, hosted runtime behavior, or generated reports. * Historical hackathon language is retained only where intentionally historical. ### Group 9 - ADRs, API, And Planning Records Goal: Audit low-churn records separately so planning docs do not silently become current-behavior docs. Files: * `docs/adr/0000-template.md` * `docs/adr/0001-extension-platform-foundation.md` * `docs/api/README_api.md` * `docs/ongoing-projects/sqlite-observation-store-transition-plan.md` Retired records deleted in the current worktree: * `docs/ongoing-projects/engine-replay-reference-tabs-plan.md` * `docs/ongoing-projects/skill-saved-time-file-overrides-plan.md` * `docs/ongoing-projects/trend-finder-card-visibility-plan.md` These implemented-plan records are no longer active docs unless a future session intentionally restores them. Verification sources: * `src/routes/` * `src/lib/` * `scripts/` * Any referenced issue, phase, or implementation file in the plan itself Exit criteria: * ADRs remain decision records, not mutable product docs. * Ongoing plans clearly show status, owner assumptions, and whether work is implemented, active, or deferred. ### Group 10 - Directory READMEs Goal: Verify each directory-level README explains local ownership without duplicating root docs. Files: * `data/README_data.md` * `logs/README_logs.md` * `public/README_public.md` * `public/assets/trend-finder/README.md` * `scripts/README_scripts.md` * `scripts/extensions/README_extensions-scripts.md` * `skills/README_skills.md` * `src/README_src.md` * `src/extensions/README_extensions.md` * `tests/README_tests.md` Verification sources: * The directory containing each README * `package.json` * `.gitignore` Exit criteria: * Subdirectory README naming follows the repo convention where practical. * Each README points to owner docs instead of duplicating long explanations. * Public assets, generated data, logs, scripts, skills, source, and tests are described accurately. ### Group 11 - Local Skill Docs Goal: Audit repo-local skill files as operational artifacts, not product docs. Files: * `skills/dream/SKILL.md` * `skills/personas/SKILL.md` Verification sources: * `skills/README_skills.md` * `scripts/install-dream-cron.ts` * Any script or route that references local skills Exit criteria: * Skill instructions match current runtime behavior. * Any local/private data references preserve read-only and privacy boundaries. ### Group 12 - Example And Imported Reference Docs Goal: Decide whether ignored imported examples remain useful, need archival labels, or should be excluded from the active docs set. Disposition: * `EXAMPLES/` is an ignored imported reference snapshot, not active AI OS documentation. * Do not maintain, format, lint, or test these imported docs during normal docs audits. * Preserve imported license files and upstream example content unless a future session intentionally promotes a specific example into tracked documentation. Files: * `EXAMPLES/claudeos/DIFF_V1_V2.md` * `EXAMPLES/claudeos/claude-os-v1/CLAUDE.md` * `EXAMPLES/claudeos/claude-os-v1/LICENSE` * `EXAMPLES/claudeos/claude-os-v1/README.md` * `EXAMPLES/claudeos/claude-os-v1/skills/dream/SKILL.md` * `EXAMPLES/claudeos/claude-os-v2/ClaudeOS [Hermes] V2/CLAUDE.md` * `EXAMPLES/claudeos/claude-os-v2/ClaudeOS [Hermes] V2/LICENSE` * `EXAMPLES/claudeos/claude-os-v2/ClaudeOS [Hermes] V2/README.md` * `EXAMPLES/claudeos/claude-os-v2/ClaudeOS [Hermes] V2/skills/dream/SKILL.md` * `EXAMPLES/claudeos/claude-os-v2/ClaudeOS [Hermes] V2/skills/personas/SKILL.md` Verification sources: * `docs/project-identity.md` * `docs/PRD.md` * Any current docs that reference `EXAMPLES/` Exit criteria: * Imported docs are labeled as reference/historical if retained. * Current AI OS docs do not depend on stale imported instructions. * License references for imported examples remain intact. ### Group 13 - Generated Or Ignored Audit Artifacts Goal: Treat generated audit reports as evidence snapshots, not maintained product docs. Disposition: * `dogfood-output/` is ignored generated browser-audit output. * Keep the 2026-05-25 Trend Finder dogfood report as local evidence only. * Do not maintain, format, lint, or test generated dogfood output during normal documentation audits. Files: * `dogfood-output/trend-finder-20260525-090500/report.md` Verification sources: * `docs/hackathon/trend-finder-demo.md` * `docs/testing.md` * `tests/README_tests.md` Exit criteria: * The report is either explicitly retained as a snapshot or excluded from active docs maintenance. * No current docs depend on generated report claims without a current source. ## Suggested Pass Order 1. Group 0: refresh inventory and audit control notes. 2. Group 2: lock product identity and source-of-truth claims. 3. Group 1: sync root docs from the source-of-truth set. 4. Group 3: validate setup, development, deployment, and testing docs. 5. Group 4: validate host runtime and command docs. 6. Groups 5-6: validate Trend Finder extension, scoring, and source docs. 7. Group 7: validate runbooks and private artifact boundaries. 8. Group 8: validate public demo and research docs. 9. Group 9: validate ADRs and planning records. 10. Groups 10-11: validate directory READMEs and local skills. 11. Groups 12-13: disposition imported and generated docs. 12. Group 0 again: update this file and `docs/CHANGELOG.md` with the final audit outcome. ## Validation Expectations For docs-only edits: ```bash bun run format:check ``` For docs that change commands, routes, environment behavior, runtime contracts, or user-visible claims, add the narrowest relevant checks: ```bash bun run typecheck bun run test ``` Use `bun run test:e2e` only when route behavior, setup flows, browser-visible documentation surfaces, or demos are changed. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://ai-os-and-trend-finder.gitbook.io/ai-os-and-trend-finder-docs/docs/docs-audit.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.