> 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/.spec_system/archive/sessions/phase31-session04-extensions-and-agent-routes/implementation-notes.md).

# Implementation Notes

**Session ID**: `phase31-session04-extensions-and-agent-routes` **Started**: 2026-06-24 12:15 **Last Updated**: 2026-06-24 12:31

***

## Session Progress

| Metric              | Value     |
| ------------------- | --------- |
| Tasks Completed     | 15 / 21   |
| Estimated Remaining | 3-4 hours |
| Blockers            | 0         |

***

## Task Log

### 2026-06-24 - Session Start

**Environment verified**:

* [x] Prerequisites confirmed
* [x] Tools available
* [x] Directory structure ready

***

### Task T001 - Verify Phase 31 prerequisites and Session 04 scope

**Started**: 2026-06-24 12:15 **Completed**: 2026-06-24 12:15 **Duration**: 1 minute

**Notes**:

* Verified Phase 31 PRD progress tracker records Sessions 01 through 03 complete and Session 04 as the next unfinished session.
* Confirmed Session 04 scope covers public-demo route behavior for agent routes, Knowledge Graph, Trend Finder, and AI Rogue, with Pages build output deferred to Session 05.

**Files Changed**:

* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Created session progress log and recorded T001 evidence.

**Verification**:

* Command/check: `sed -n '1,260p' .spec_system/PRD/phase_31/PRD_phase_31.md && sed -n '261,560p' .spec_system/PRD/phase_31/PRD_phase_31.md`
  * Result: PASS - Phase 31 PRD shows Sessions 01-03 complete, Session 04 not started, and the expected public demo route scope.
  * Evidence: Read Phase 31 Progress Tracker, Demo Build Mode, Feature Matrix, and Implementation Sessions sections.
* UI product-surface check: N/A - Documentation and planning artifact review only.
* UI craft check: N/A - No UI changed.

***

### Task T002 - Review Session 04 endpoint inventory

**Started**: 2026-06-24 12:15 **Completed**: 2026-06-24 12:15 **Duration**: 1 minute

**Notes**:

* Reviewed the Session 04 endpoint inventory and route behavior matrix.
* Confirmed implementation must avoid Trend Finder, Knowledge Graph, Hermes, and OpenClaw bridge endpoints in public-demo mode while keeping routes visible with demo, read-only, unavailable, or browser-local states.

**Files Changed**:

* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Added T002 endpoint inventory evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated checklist progress for completed setup tasks.

**Verification**:

* Command/check: `sed -n '1,260p' .spec_system/PRD/phase_31/session_04_extensions_and_agent_routes.md`
  * Result: PASS - Session 04 artifact lists endpoint groups for Trend Finder, Knowledge Graph, Hermes, and OpenClaw plus route behavior expectations for registered extensions and agents.
  * Evidence: Read Folded Source Details, route behavior matrix, prerequisites, deliverables, and success criteria.
* UI product-surface check: N/A - Documentation and planning artifact review only.
* UI craft check: N/A - No UI changed.

***

### Task T003 - Create public-demo route fetch recorder

**Started**: 2026-06-24 12:16 **Completed**: 2026-06-24 12:17 **Duration**: 1 minute

**Notes**:

* Added a reusable public-demo fetch recorder that stores all requested paths and records bridge paths beginning with `/__`.
* The recorder throws on local bridge paths so route tests fail at the first unsafe public-demo request.

**Files Changed**:

* `src/routes/__tests__/public-demo-routes.test.tsx` - Created fetch recorder helpers and focused helper coverage.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T003 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/public-demo-routes.test.tsx`
  * Result: PASS - 1 test file passed, 2 tests passed.
  * Evidence: Helper records `/demo/live-data.snapshot.json` as allowed and rejects `/__token` as a local bridge path.
* UI product-surface check: N/A - Test helper only.
* UI craft check: N/A - No UI changed.

**BQC Fixes**:

* Trust boundary enforcement: The helper treats any route request starting with `/__` as a public-demo trust-boundary violation and fails the test immediately (`src/routes/__tests__/public-demo-routes.test.tsx`).

***

### Task T004 - Add hosted-demo copy keys

**Started**: 2026-06-24 12:17 **Completed**: 2026-06-24 12:17 **Duration**: 1 minute

**Notes**:

* Added shared public-demo copy keys for agent routes, AI Rogue browser-local state, Knowledge Graph demo fixtures, and Trend Finder reviewed media.
* Extended public-demo helper tests to ensure route copy is product-facing and does not expose internal bridge endpoint names.

**Files Changed**:

* `src/lib/public-demo.ts` - Added hosted route, graph, media, and browser-local copy entries.
* `src/lib/__tests__/public-demo.test.ts` - Added copy contract coverage for the new keys.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T004 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/lib/__tests__/public-demo.test.ts`
  * Result: PASS - 1 test file passed, 27 tests passed.
  * Evidence: New copy keys return non-empty product-facing labels and avoid `/__` endpoint strings in title/message copy.
* UI product-surface check: PASS - Copy is product-facing hosted demo language and does not name bridge endpoints.
* UI craft check: N/A - Shared copy only; no layout changed.

**BQC Fixes**:

* Product surface discipline: Centralized hosted-demo copy for route surfaces so components can avoid diagnostic endpoint names (`src/lib/public-demo.ts`).

***

### Task T005 - Add Trend Finder static asset preview coverage

**Started**: 2026-06-24 12:18 **Completed**: 2026-06-24 12:19 **Duration**: 1 minute

**Notes**:

* Added focused evidence asset preview tests for reviewed static public image/video URLs under `/demo/trend-finder-assets/*`.
* Added the narrow schema, view-model, and component support needed for the tests to pass without token or bridge fetches.

**Files Changed**:

* `src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx` - Created focused static asset/no-token tests.
* `src/extensions/trend-finder/schema.ts` - Allowed reviewed public demo asset paths through the browser-safe asset URL schema.
* `src/extensions/trend-finder/view-model.ts` - Split static public asset URLs from local bridge asset URLs in the preview view model.
* `src/extensions/trend-finder/components/evidence-asset-preview.tsx` - Rendered static public media directly without token fetches.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T005 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx`
  * Result: PASS - 1 test file passed, 3 tests passed.
  * Evidence: Static image and video previews render from `/demo/trend-finder-assets/*` and the fetch mock is not called.
* UI product-surface check: PASS - Static media renders as product evidence previews; unavailable states use fallback copy without endpoint names.
* UI craft check: PASS - Existing compact preview dimensions, icons, alt text, and focus behavior are preserved.

**BQC Fixes**:

* Trust boundary enforcement: Static public asset URLs are allowlisted and separated from token-gated local bridge URLs (`src/extensions/trend-finder/schema.ts`, `src/extensions/trend-finder/view-model.ts`).
* Resource cleanup: The preview component revokes prior object URLs when switching away from local bridge previews and skips scoped fetch resources for public assets (`src/extensions/trend-finder/components/evidence-asset-preview.tsx`).

***

### Task T006 - Add AI Rogue public-demo expectations

**Started**: 2026-06-24 12:19 **Completed**: 2026-06-24 12:20 **Duration**: 1 minute

**Notes**:

* Added AI Rogue public-demo test coverage proving the extension remains visible even when normal extension env disables it.
* Updated AI Rogue shell status copy so hosted public-demo routes present browser-local persistence and disable the LiveData refresh affordance.

**Files Changed**:

* `src/extensions/ai-rogue/__tests__/client.test.tsx` - Added public-demo extension visibility and browser-local shell assertions.
* `src/extensions/ai-rogue/views/view-shell.tsx` - Added public-demo browser-local status copy and disabled refresh affordance.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T006 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/extensions/ai-rogue/__tests__/client.test.tsx`
  * Result: PASS - 1 test file passed, 19 tests passed.
  * Evidence: Public demo mode keeps `ai-rogue` in `getEnabledExtensions()` and the Play shell shows browser-local hosted-demo copy with refresh disabled.
* UI product-surface check: PASS - Public-demo AI Rogue route copy says browser-local storage and does not expose bridge endpoint names.
* UI craft check: PASS - Existing compact shell status/banner layout is preserved while the icon button gets an accurate disabled label/title.

**BQC Fixes**:

* Duplicate action prevention: Public-demo refresh is disabled in the shell so users cannot trigger misleading hosted refresh actions (`src/extensions/ai-rogue/views/view-shell.tsx`).
* Product surface discipline: Public-demo AI Rogue status uses product-facing browser-local copy rather than raw runtime status labels (`src/extensions/ai-rogue/views/view-shell.tsx`).

***

## Checkpoints

### Checkpoint 1 - Foundation Complete

**Time**: 2026-06-24 12:20 **Progress**: 6 / 21 tasks complete. **Current scope**: Setup and foundation tasks are complete. Public-demo helper, shared route copy, Trend Finder static asset preview coverage, and AI Rogue browser-local public-demo expectations are in place. **Next task**: T007, Hermes hosted public-demo route coercion.

***

### Task T007 - Force Hermes hosted public-demo mode

**Started**: 2026-06-24 12:20 **Completed**: 2026-06-24 12:22 **Duration**: 2 minutes

**Notes**:

* Coerced Hermes route `demoMode` from `VITE_AI_OS_PUBLIC_DEMO` before `useHermes` and admin hooks initialize.
* Reset selected chat session state on hosted-demo entry and prevented return-to-live controls from rendering in public-demo mode.
* Added route coverage proving hosted public-demo Hermes renders sample data without any fetch calls.

**Files Changed**:

* `src/components/hermes/hermes-read-only-page.tsx` - Forced hosted demo state, reset selected session state, and hid live-mode toggles in public-demo mode.
* `src/components/hermes/hermes-status-bar.tsx` - Added `canUseLive` control so hosted-demo mode can suppress return-to-live actions.
* `src/routes/__tests__/agents.test.tsx` - Added Hermes public-demo no-fetch route assertion.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T007 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/agents.test.tsx`
  * Result: PASS - 1 test file passed, 30 tests passed.
  * Evidence: New Hermes public-demo test stubs `VITE_AI_OS_PUBLIC_DEMO=1`, renders sample session activity, finds hosted-demo copy, finds no return-to-live button, and verifies `fetch` was not called.
* UI product-surface check: PASS - Hosted-demo Hermes shows product-facing sample-data copy and no bridge endpoint names or return-to-live controls.
* UI craft check: PASS - Existing Hermes shell, banner, status bar, tabs, and demo styling are preserved.

**BQC Fixes**:

* State freshness on re-entry: Hosted-demo entry resets selected chat session state so stale live session detail cannot carry into demo mode (`src/components/hermes/hermes-read-only-page.tsx`).
* Trust boundary enforcement: Public-demo state is calculated before `useHermes` and `useHermesAdmin` receive options, preventing bridge query construction (`src/components/hermes/hermes-read-only-page.tsx`).

***

### Task T008 - Force Claude Code hosted public-demo behavior

**Started**: 2026-06-24 12:22 **Completed**: 2026-06-24 12:23 **Duration**: 1 minute

**Notes**:

* Coerced Claude Code Mission Control into demo mode from `VITE_AI_OS_PUBLIC_DEMO` before Hermes read/admin hooks initialize.
* Suppressed Live/Demo and return-to-live controls in hosted public-demo mode while preserving normal local controls outside public demo.
* Added route coverage proving Claude Code renders bundled demo missions without bridge fetches.

**Files Changed**:

* `src/components/hermes/claude-code-mission-page.tsx` - Forced hosted demo mode and hid live-mode controls in public-demo mode.
* `src/routes/__tests__/agents.test.tsx` - Added Claude Code public-demo no-fetch route assertion.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T008 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/agents.test.tsx`
  * Result: PASS - 1 test file passed, 31 tests passed.
  * Evidence: New Claude Code public-demo test stubs `VITE_AI_OS_PUBLIC_DEMO=1`, renders demo mission data, finds hosted-demo copy, verifies live controls are absent, and verifies `fetch` was not called.
* UI product-surface check: PASS - Hosted-demo Claude Code shows product-facing sample-data copy and no return-to-live/live-mode controls.
* UI craft check: PASS - Existing Claude Code shell, hero, mission surface, and responsive controls are preserved.

**BQC Fixes**:

* Trust boundary enforcement: Public-demo state is calculated before Hermes read/admin hooks receive options, preventing Claude Code route bridge query construction (`src/components/hermes/claude-code-mission-page.tsx`).
* Product surface discipline: Hosted demo removes live-mode controls that would imply local process or admin availability (`src/components/hermes/claude-code-mission-page.tsx`).

***

### Task T009 - Force OpenClaw hosted public-demo mode

**Started**: 2026-06-24 12:23 **Completed**: 2026-06-24 12:24 **Duration**: 1 minute

**Notes**:

* Coerced OpenClaw route demo mode from `VITE_AI_OS_PUBLIC_DEMO` before status, token, dashboard, and admin hooks initialize.
* Reset selected tab/mode state on hosted-demo entry and suppressed return-to-live controls in public-demo mode.
* Added route coverage proving OpenClaw public-demo mode renders bundled demo data without bridge fetches.

**Files Changed**:

* `src/components/openclaw/openclaw-read-only-page.tsx` - Forced hosted demo state, reset selected route state, and passed public-demo control flags.
* `src/components/openclaw/openclaw-status-bar.tsx` - Added `canUseLive` control to suppress return-to-live actions.
* `src/routes/__tests__/agents.test.tsx` - Added OpenClaw public-demo no-fetch route assertion.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T009 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/agents.test.tsx`
  * Result: PASS - 1 test file passed, 32 tests passed.
  * Evidence: New OpenClaw public-demo test stubs `VITE_AI_OS_PUBLIC_DEMO=1`, renders demo swarm data, finds hosted-demo copy, verifies return-to-live controls are absent, and verifies `fetch` was not called.
* UI product-surface check: PASS - Hosted-demo OpenClaw shows product-facing sample-data copy and no return-to-live control.
* UI craft check: PASS - Existing OpenClaw shell, status bar, tabs, and demo visual treatment are preserved.

**BQC Fixes**:

* State freshness on re-entry: Hosted-demo entry resets selected OpenClaw mode and returns the page to the Swarms tab (`src/components/openclaw/openclaw-read-only-page.tsx`).
* Trust boundary enforcement: Public-demo state is calculated before OpenClaw read/admin hooks receive options, preventing bridge query construction (`src/components/openclaw/openclaw-read-only-page.tsx`).

***

### Task T010 - Force Knowledge Graph public-demo fixtures through route wiring

**Started**: 2026-06-24 12:25 **Completed**: 2026-06-24 12:26 **Duration**: 1 minute

**Notes**:

* Passed `VITE_AI_OS_PUBLIC_DEMO` into `useKnowledgeGraph()` at the route page boundary before registry, graph, token, and admin hook paths initialize.
* Updated Knowledge Graph demo status detail to use shared hosted-demo product copy instead of the older internal fixture notice.
* Added public-demo route coverage proving the page renders demo graph state without bridge requests.

**Files Changed**:

* `src/components/knowledge-graph/knowledge-graph-page.tsx` - Passed public-demo mode into the Knowledge Graph hook.
* `src/hooks/use-knowledge-graph.ts` - Used product-facing hosted-demo copy for demo status detail.
* `src/routes/__tests__/public-demo-routes.test.tsx` - Added Knowledge Graph public-demo no-bridge route assertion.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T010 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/public-demo-routes.test.tsx`
  * Result: PASS - 1 test file passed, 3 tests passed.
  * Evidence: Knowledge Graph public-demo route renders `Demo graph`, shows hosted-demo graph copy, and the fetch recorder records no `/__*` or other fetch requests.
* UI product-surface check: PASS - Demo status detail uses product-facing local graph service copy and avoids endpoint names or internal binary labels.
* UI craft check: PASS - Existing Knowledge Graph hero, gallery, and panel structure are preserved.

**BQC Fixes**:

* Trust boundary enforcement: Public-demo mode is passed at the route page boundary before graph registry, token, admin, and Hermes projection hooks can construct bridge queries (`src/components/knowledge-graph/knowledge-graph-page.tsx`).
* Product surface discipline: Public demo Knowledge Graph status copy no longer surfaces the internal fixture notice on the primary page (`src/hooks/use-knowledge-graph.ts`).

***

### Checkpoint 2 - Agent And Graph Boundaries

**Time**: 2026-06-24 12:26 **Progress**: 10 / 21 tasks complete. **Current scope**: Hermes, Claude Code, OpenClaw, and Knowledge Graph page boundaries are now forced into hosted-demo mode with focused no-fetch route coverage. **Next task**: T011, grounded Hermes chat read-only behavior for public-demo Knowledge Graph.

***

### Task T011 - Keep grounded Hermes chat demo/read-only in public-demo mode

**Started**: 2026-06-24 12:26 **Completed**: 2026-06-24 12:27 **Duration**: 1 minute

**Notes**:

* Added a defensive browser public-demo check inside `KnowledgeGraphGroundedChat` so Hermes read/admin hooks stay demo/read-only even if a parent forgets to pass demo mode.
* Extended route coverage to assert grounded Hermes chat renders read-only demo copy and makes no fetches.

**Files Changed**:

* `src/components/knowledge-graph/knowledge-graph-grounded-chat.tsx` - Added explicit public-demo guard and passed effective demo mode to Hermes chat.
* `src/routes/__tests__/public-demo-routes.test.tsx` - Added grounded chat read-only public-demo assertions.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T011 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/public-demo-routes.test.tsx`
  * Result: PASS - 1 test file passed, 3 tests passed.
  * Evidence: Knowledge Graph public-demo route renders `Grounded Hermes` with read-only demo copy and the fetch recorder records no requests.
* UI product-surface check: PASS - Grounded chat shows product-facing read-only demo copy and no endpoint names.
* UI craft check: PASS - Existing grounded chat panel structure and status chip behavior are preserved.

**BQC Fixes**:

* Trust boundary enforcement: Grounded chat now computes an effective demo mode from both the Knowledge Graph view and browser public-demo flag before initializing Hermes read/admin hooks (`src/components/knowledge-graph/knowledge-graph-grounded-chat.tsx`).
* State freshness on re-entry: Selected grounded chat session resets when effective demo mode or active project changes (`src/components/knowledge-graph/knowledge-graph-grounded-chat.tsx`).

***

### Task T012 - Skip Trend Finder visibility file bridge in public-demo mode

**Started**: 2026-06-24 12:27 **Completed**: 2026-06-24 12:28 **Duration**: 1 minute

**Notes**:

* Updated the Trend Finder visibility provider to hydrate from localStorage and mark itself hydrated without fetching the visibility file bridge in public-demo mode.
* Added focused provider coverage proving public-demo visibility keeps browser-local hidden items and does not call fetch.

**Files Changed**:

* `src/extensions/trend-finder/use-visibility-settings.ts` - Added public-demo no-file-fetch branch with re-entry hydration reset.
* `src/extensions/trend-finder/__tests__/visibility-views.test.tsx` - Added public-demo localStorage/no-fetch visibility coverage.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T012 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/extensions/trend-finder/__tests__/visibility-views.test.tsx`
  * Result: PASS - 1 test file passed, 12 tests passed.
  * Evidence: Public-demo visibility provider reads `topic:topic-001` from localStorage, reports hydrated, and the fetch mock is not called.
* UI product-surface check: N/A - Visibility state provider change; no visible copy changed.
* UI craft check: N/A - No layout changed.

**BQC Fixes**:

* State freshness on re-entry: Visibility hydration resets when public-demo mode changes so stale file-merged state cannot persist across mode boundaries (`src/extensions/trend-finder/use-visibility-settings.ts`).
* Trust boundary enforcement: Public-demo mode returns before `TREND_FINDER_VISIBILITY_CONFIG_ENDPOINT` can be fetched (`src/extensions/trend-finder/use-visibility-settings.ts`).

***

### Task T013 - Allow reviewed static Trend Finder asset URLs in schema

**Started**: 2026-06-24 12:28 **Completed**: 2026-06-24 12:28 **Duration**: 1 minute

**Notes**:

* Added a narrow `/demo/trend-finder-assets/*` public asset allowlist alongside the existing local asset bridge path.
* Added schema tests proving reviewed public paths survive parsing while traversal, query-bearing, remote, and unrelated demo paths are cleared.

**Files Changed**:

* `src/extensions/trend-finder/schema.ts` - Added `isTrendFinderPublicAssetUrl()` and updated browser asset URL parsing.
* `src/lib/__tests__/trend-finder-schema.test.ts` - Added static public asset allowlist and unsafe URL rejection assertions.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T013 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/lib/__tests__/trend-finder-schema.test.ts`
  * Result: PASS - 1 test file passed, 115 tests passed.
  * Evidence: Reviewed `/demo/trend-finder-assets/reviewed-thumbnail.webp` survives parsing; unsafe traversal, duplicate slash, query, remote, and unrelated demo URLs are cleared.
* UI product-surface check: N/A - Schema parsing only.
* UI craft check: N/A - No UI changed.

**BQC Fixes**:

* Trust boundary enforcement: Public demo media URLs are allowlist-first and reject arbitrary local, remote, traversal, query-bearing, and unrelated paths (`src/extensions/trend-finder/schema.ts`).

***

### Task T014 - Split Trend Finder evidence asset view models

**Started**: 2026-06-24 12:29 **Completed**: 2026-06-24 12:30 **Duration**: 1 minute

**Notes**:

* Split evidence asset preview view models into `publicAssetUrl` and `localBridgeUrl` fields.
* Added view-model coverage for public static assets, protected local bridge assets, and unavailable fallback states.

**Files Changed**:

* `src/extensions/trend-finder/view-model.ts` - Added public/static asset URL split and public asset flag in preview view models.
* `src/extensions/trend-finder/__tests__/view-model.test.ts` - Added evidence asset preview split assertions.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T014 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/extensions/trend-finder/__tests__/view-model.test.ts`
  * Result: PASS - 1 test file passed, 65 tests passed.
  * Evidence: View-model test confirms public demo assets populate `publicAssetUrl` only, bridge assets populate `localBridgeUrl` only, and unavailable assets keep fallback labels without renderable URLs.
* UI product-surface check: N/A - View model projection only.
* UI craft check: N/A - No layout changed.

**BQC Fixes**:

* Contract alignment: Preview view models now expose explicit fields for static public assets and local bridge assets, matching schema and component behavior (`src/extensions/trend-finder/view-model.ts`).
* Trust boundary enforcement: Public and bridge asset paths are mutually exclusive in the rendered view model (`src/extensions/trend-finder/view-model.ts`).

***

### Task T015 - Render static Trend Finder asset previews directly

**Started**: 2026-06-24 12:30 **Completed**: 2026-06-24 12:31 **Duration**: 1 minute

**Notes**:

* Updated evidence asset previews to render static `/demo/trend-finder-assets/*` media directly without token or bridge fetches.
* Preserved protected local bridge preview behavior behind `/__token` and added cleanup coverage for object URLs.

**Files Changed**:

* `src/extensions/trend-finder/components/evidence-asset-preview.tsx` - Rendered public asset URLs directly and skipped scoped local fetches for public media.
* `src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx` - Added static media/no-token and token-gated bridge cleanup coverage.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T015 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx`
  * Result: PASS - 1 test file passed, 4 tests passed.
  * Evidence: Public image/video media render from static URLs without fetch; local bridge preview fetches `/__token` before `/__trend_finder_asset/*` and revokes the object URL on unmount.
* UI product-surface check: PASS - Static media renders as evidence preview content; unavailable states show fallback copy without endpoint names.
* UI craft check: PASS - Existing preview size, alt text, focus behavior, and icon fallback styling are preserved.

**BQC Fixes**:

* Resource cleanup: Local bridge object URLs are revoked on scope exit and when public/unavailable assets replace bridge previews (`src/extensions/trend-finder/components/evidence-asset-preview.tsx`).
* Trust boundary enforcement: Static public media never uses token fetches, while protected bridge media remains token-gated outside public-demo static paths (`src/extensions/trend-finder/components/evidence-asset-preview.tsx`).

***

### Checkpoint 3 - Trend Finder Media And Visibility

**Time**: 2026-06-24 12:31 **Progress**: 15 / 21 tasks complete. **Current scope**: Trend Finder visibility settings skip file bridges in public demo, and evidence assets now have schema, view-model, component, and test coverage for static public media vs protected bridge media. **Next task**: T016, Trend Finder route provenance/unavailable-state copy.

***

### Task T016 - Tune Trend Finder public-demo provenance labels

**Started**: 2026-06-24 12:31 **Completed**: 2026-06-24 12:34 **Duration**: 3 minutes

**Notes**:

* Added product-facing public-demo snapshot notices to Trends, Sources, and Engine Replay.
* Renamed visible Trend Finder diagnostics copy to readiness/details language while keeping existing visibility IDs stable.
* Made Sources explicitly disable source setup and scheduler auto-load in public-demo mode before hook guards run.
* Cleaned Engine Replay stage and asset descriptions so public-facing replay copy no longer implies raw logs, environmental tokens, endpoint details, or protected bridge access.

**Files Changed**:

* `src/extensions/trend-finder/views/trends-view.tsx` - Added snapshot notice and renamed the controls surface to readiness copy.
* `src/extensions/trend-finder/views/sources-view.tsx` - Added source/scheduler public-demo notice and public-demo auto-load guard.
* `src/extensions/trend-finder/views/engine-replay-view.tsx` - Added public-demo replay snapshot/reviewed-media notice.
* `src/extensions/trend-finder/components/runtime-readiness-panel.tsx` - Renamed the details disclosure from diagnostics to runtime details.
* `src/extensions/trend-finder/engine-replay-model.ts` - Reworded replay stage and asset descriptions away from raw/bridge implications.
* `src/lib/__tests__/trend-finder-dashboard.test.tsx` - Added public-demo source/scheduler no-bridge coverage and updated readiness copy assertions.
* `src/lib/__tests__/trend-finder-engine-replay.test.tsx` - Added public-demo replay notice coverage.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T016 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/lib/__tests__/trend-finder-dashboard.test.tsx`
  * Result: PASS - 1 test file passed, 36 tests passed.
  * Evidence: Public-demo Trends shows frozen snapshot copy; Sources shows demo source/scheduler labels and makes no `/__*` setup or scheduler request.
* Command/check: `bun run test -- src/lib/__tests__/trend-finder-engine-replay.test.tsx`
  * Result: PASS - 1 test file passed, 23 tests passed.
  * Evidence: Engine Replay public-demo notice labels replay state as demo snapshot plus reviewed media using sanitized public snapshot copy.
* UI product-surface check: PASS - Visible copy now uses snapshot, readiness, reviewed media, and browser-safe labels without endpoint names.
* UI craft check: PASS - Notices use existing compact card styles and do not alter fixed replay/Trend Finder layouts.

**BQC Fixes**:

* Product-copy boundary: Replaced user-visible diagnostics wording on the primary Trend Finder surface with readiness/details language (`src/extensions/trend-finder/views/trends-view.tsx`, `src/extensions/trend-finder/components/runtime-readiness-panel.tsx`).
* Trust boundary enforcement: Sources view now passes `autoLoad: false` for source setup and scheduler controls in public-demo mode before bridge-backed hooks can load (`src/extensions/trend-finder/views/sources-view.tsx`).

***

### Checkpoint 4 - Trend Finder Public Demo Copy

**Time**: 2026-06-24 12:34 **Progress**: 16 / 21 tasks complete. **Current scope**: Trend Finder route labels, source/scheduler unavailable states, and Engine Replay copy are public-demo product-facing with focused no-bridge coverage. **Next task**: T017, AI Rogue browser-local Play and Ledger labeling.

***

### Task T017 - Label AI Rogue public-demo state as browser-local

**Started**: 2026-06-24 12:34 **Completed**: 2026-06-24 12:36 **Duration**: 2 minutes

**Notes**:

* Added explicit browser-local public-demo run-state copy to Play while keeping runtime save/load/reset handlers unchanged.
* Renamed the Play collapsible details surface from developer diagnostics to product-facing run tools.
* Added Ledger public-demo claim copy that states claim records stay in this browser and do not create hosted server state.
* Added focused Ledger coverage that double-clicking Claim in public-demo mode still records exactly one durable claim.

**Files Changed**:

* `src/extensions/ai-rogue/views/play-view.tsx` - Added Browser-local run state panel and public-demo runtime-unavailable copy.
* `src/extensions/ai-rogue/views/ledger-view.tsx` - Added public-demo browser-local claim labeling.
* `src/extensions/ai-rogue/__tests__/client.test.tsx` - Added Play copy assertions and public-demo Ledger idempotent-claim coverage.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T017 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/extensions/ai-rogue/__tests__/client.test.tsx`
  * Result: PASS - 1 test file passed, 20 tests passed.
  * Evidence: Public-demo Play renders Browser local state and disabled refresh; public-demo Ledger renders browser-local claim copy and a rapid double Claim click calls `writeAiRogueClaimEntryDurable` once.
* UI product-surface check: PASS - Public-demo AI Rogue labels consistently say browser-local and avoid hosted server persistence language.
* UI craft check: PASS - New Play panel reuses existing compact panel/signal-list layout without changing runtime canvas dimensions.

**BQC Fixes**:

* Duplicate-trigger protection: Added regression coverage for the Ledger `claimInFlightRef` guard in public-demo mode (`src/extensions/ai-rogue/__tests__/client.test.tsx`).
* Product-copy boundary: Replaced Play developer-diagnostics label with run-tools copy and added explicit browser-local state labels for public demo (`src/extensions/ai-rogue/views/play-view.tsx`, `src/extensions/ai-rogue/views/ledger-view.tsx`).

***

### Checkpoint 5 - Implementation Tasks Complete

**Time**: 2026-06-24 12:36 **Progress**: 17 / 21 tasks complete. **Current scope**: Agent, Knowledge Graph, Trend Finder, and AI Rogue implementation tasks are complete. Remaining work is route-test expansion plus focused validation commands. **Next task**: T018, agent public-demo route no-bridge tests.

***

### Task T018 - Extend agent route public-demo no-bridge tests

**Started**: 2026-06-24 12:36 **Completed**: 2026-06-24 12:37 **Duration**: 1 minute

**Notes**:

* Confirmed the agent route suite includes hosted public-demo cases for Claude Code, Hermes, and OpenClaw.
* Each public-demo test stubs `VITE_AI_OS_PUBLIC_DEMO`, renders demo/sample data, hides live/return affordances, and asserts the fetch mock is not called.

**Files Changed**:

* `src/routes/__tests__/agents.test.tsx` - Existing public-demo no-bridge route tests verified for Hermes, Claude Code, and OpenClaw.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T018 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/agents.test.tsx`
  * Result: PASS - 1 test file passed, 32 tests passed.
  * Evidence: Agent route suite includes public-demo tests where Claude Code, Hermes, and OpenClaw render demo data and `fetch` receives no bridge calls.
* UI product-surface check: PASS - Hosted demo agent routes show shared unavailable copy and hide live return controls.
* UI craft check: PASS - Route skins remain scoped to Hermes, Claude Code, and OpenClaw wrappers.

**BQC Fixes**:

* Regression coverage: Agent route tests now guard public-demo mode against Hermes/OpenClaw bridge fetches and Claude-specific bridge expansion (`src/routes/__tests__/agents.test.tsx`).

***

### Checkpoint 6 - Agent Route Tests

**Time**: 2026-06-24 12:37 **Progress**: 18 / 21 tasks complete. **Current scope**: Agent route public-demo no-bridge coverage is passing. **Next task**: T019, Knowledge Graph and extension public-demo route no-bridge tests.

***

### Task T019 - Extend Knowledge Graph and extension public-demo no-bridge tests

**Started**: 2026-06-24 12:37 **Completed**: 2026-06-24 12:39 **Duration**: 2 minutes

**Notes**:

* Switched the Knowledge Graph route test to render the real page with only the 3D graph component mocked, then added a public-demo no-bridge assertion.
* Added extension route public-demo fetch recorders that allow `/demo/live-data.snapshot.json` and fail on any `/__*` bridge path.
* Added public-demo route coverage for Trend Finder Trends, Trend Finder Engine Replay, and AI Rogue Play.
* The new extension route recorder exposed a real `/__token` request from protected Trend Finder evidence assets in public-demo mode; fixed the asset preview guard so local bridge previews never fetch in public-demo mode.

**Files Changed**:

* `src/routes/__tests__/knowledge-graph.test.tsx` - Added real Knowledge Graph public-demo route no-bridge coverage.
* `src/routes/__tests__/extensions-routes.test.tsx` - Added public-demo no-bridge coverage for Trend Finder and AI Rogue extension routes.
* `src/extensions/trend-finder/components/evidence-asset-preview.tsx` - Added public-demo no-bridge guard for protected local asset previews.
* `src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx` - Added regression coverage for public-demo local bridge assets.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T019 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/knowledge-graph.test.tsx`
  * Result: PASS - 1 test file passed, 3 tests passed.
  * Evidence: Knowledge Graph route renders demo graph and grounded chat copy in public-demo mode with no fetch calls.
* Command/check: `bun run test -- src/routes/__tests__/extensions-routes.test.tsx`
  * Result: PASS - 1 test file passed, 20 tests passed.
  * Evidence: Trend Finder, Engine Replay, and AI Rogue public-demo extension routes may fetch `/demo/live-data.snapshot.json` but record no `/__*` bridge paths.
* Command/check: `bun run test -- src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx`
  * Result: PASS - 1 test file passed, 5 tests passed.
  * Evidence: Protected local bridge asset previews render fallback copy without `/__token` in public-demo mode.
* UI product-surface check: PASS - Public-demo route copy remains snapshot/browser-local/demo-facing.
* UI craft check: PASS - Added route tests only; asset fallback preserves existing preview dimensions.

**BQC Fixes**:

* Trust boundary enforcement: Public-demo route recorder caught and fixed a protected Trend Finder asset token fetch (`src/extensions/trend-finder/components/evidence-asset-preview.tsx`).
* Regression coverage: Knowledge Graph and extension route tests now fail immediately on public-demo `/__*` bridge requests (`src/routes/__tests__/knowledge-graph.test.tsx`, `src/routes/__tests__/extensions-routes.test.tsx`).

***

### Checkpoint 7 - Public Demo Route Matrix

**Time**: 2026-06-24 12:39 **Progress**: 19 / 21 tasks complete. **Current scope**: Agent, Knowledge Graph, Trend Finder, and AI Rogue route-level public-demo no-bridge tests are passing. **Next task**: T020, run the full focused public-demo route/component suite.

***

### Task T020 - Run focused public-demo route/component suites

**Started**: 2026-06-24 12:39 **Completed**: 2026-06-24 12:40 **Duration**: 1 minute

**Notes**:

* Ran the required focused Vitest command across public-demo route helpers, agents, Knowledge Graph, extension routes, Trend Finder asset previews, and AI Rogue client behavior.

**Files Changed**:

* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T020 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run test -- src/routes/__tests__/public-demo-routes.test.tsx src/routes/__tests__/agents.test.tsx src/routes/__tests__/knowledge-graph.test.tsx src/routes/__tests__/extensions-routes.test.tsx src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx src/extensions/ai-rogue/__tests__/client.test.tsx`
  * Result: PASS - 6 test files passed, 83 tests passed.
  * Evidence: The focused public-demo route/component matrix passed after the Trend Finder protected asset public-demo guard was added.
* UI product-surface check: PASS - Covered routes render demo/snapshot/browser-local copy.
* UI craft check: PASS - No layout regressions reported by component assertions.

**BQC Fixes**:

* Validation evidence: Required focused public-demo route and component suites are recorded with passing results.

***

### Checkpoint 8 - Focused Suite Passed

**Time**: 2026-06-24 12:40 **Progress**: 20 / 21 tasks complete. **Current scope**: Focused public-demo route/component test matrix is green. **Next task**: T021, typecheck, diff whitespace, ASCII, and LF validation.

***

### Task T021 - Run typecheck, whitespace, ASCII, and LF validation

**Started**: 2026-06-24 12:40 **Completed**: 2026-06-24 12:43 **Duration**: 3 minutes

**Notes**:

* Fixed type errors exposed by the initial validation pass:
  * Added optional chaining for a Trend Finder evidence asset fixture in the view-model test.
  * Updated the public-demo fetch recorder to use Vitest's single function-signature mock generic.
* Fixed the route-recorder-discovered public-demo asset boundary issue by ensuring protected Trend Finder local asset previews never token-fetch in public-demo mode.
* Replaced non-ASCII dash characters in the modified AI Rogue Play view so the changed-file ASCII scan is clean.
* Reran the focused public-demo route/component suite after the final type and ASCII cleanup.

**Files Changed**:

* `src/extensions/trend-finder/__tests__/view-model.test.ts` - Fixed optional evidence asset access.
* `src/routes/__tests__/public-demo-routes.test.tsx` - Fixed Vitest mock typing for the public-demo route fetch recorder.
* `src/extensions/ai-rogue/views/play-view.tsx` - Replaced non-ASCII dash copy in modified file.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/implementation-notes.md` - Recorded T021 evidence.
* `.spec_system/specs/phase31-session04-extensions-and-agent-routes/tasks.md` - Updated task progress.

**Verification**:

* Command/check: `bun run typecheck`
  * Result: PASS - `tsc --noEmit` completed without errors.
* Command/check: `bun run typecheck && git diff --check && files=$(git ls-files --modified --others --exclude-standard); if [ -n "$files" ]; then LC_ALL=C rg -n '[^\x00-\x7F]' $files; test $? -eq 1; rg -n $'\r' $files; test $? -eq 1; fi`
  * Result: PASS - typecheck and diff whitespace passed; ASCII/LF scans produced no matches.
* Command/check: `files=$(git ls-files --modified --others --exclude-standard); if [ -n "$files" ]; then LC_ALL=C rg -n '[^\x00-\x7F]' $files && exit 1 || test $? -eq 1; rg -n $'\r' $files && exit 1 || test $? -eq 1; fi`
  * Result: PASS - stricter explicit ASCII and CRLF scan produced no matches.
* Command/check: `git diff --check && files=$(git ls-files --modified --others --exclude-standard); if [ -n "$files" ]; then LC_ALL=C rg -n '[^\x00-\x7F]' $files && exit 1 || test $? -eq 1; rg -n $'\r' $files && exit 1 || test $? -eq 1; fi`
  * Result: PASS - final post-checklist whitespace, ASCII, and CRLF scan produced no matches.
* Command/check: `bun run test -- src/routes/__tests__/public-demo-routes.test.tsx src/routes/__tests__/agents.test.tsx src/routes/__tests__/knowledge-graph.test.tsx src/routes/__tests__/extensions-routes.test.tsx src/extensions/trend-finder/components/__tests__/evidence-asset-preview.test.tsx src/extensions/ai-rogue/__tests__/client.test.tsx`
  * Result: PASS - 6 test files passed, 83 tests passed.
* UI product-surface check: PASS - Public-demo copy remains snapshot/browser-local/demo-facing and changed files are ASCII-clean.
* UI craft check: PASS - No layout-affecting validation issues found.

**BQC Fixes**:

* Type safety: Public-demo recorder and Trend Finder view-model test now typecheck cleanly.
* Encoding gate: Modified files are ASCII-only with LF line endings.
* Regression confidence: Focused public-demo suite was rerun after final cleanup.

***

### Final Checkpoint - Implementation Complete

**Time**: 2026-06-24 12:43 **Progress**: 21 / 21 tasks complete. **Current scope**: Phase 31 Session 04 implementation, focused tests, typecheck, whitespace, ASCII, and LF validation are complete. **Next task**: Run `creview` to inspect and repair uncommitted changes before validation.

***


---

# 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/.spec_system/archive/sessions/phase31-session04-extensions-and-agent-routes/implementation-notes.md?ask=<question>&goal=<endgoal>
```

`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.