# Agent Review

> Aggregate all Claude sessions (Windows + WSL) with conversationId, title and description; exclude workspaces/sessions and push sessions into Task Manager

| | |
|---|---|
| Version | 1.3.2 |
| Category | AI & Agents |
| Permissions | filesystem.read, filesystem.write |
| Depends on | ai-chat |

# Agent Review

Aggregate all Claude sessions (Windows + WSL) with conversationId, title and description; exclude workspaces/sessions and push sessions into Task Manager

**Type:** `backend` — see `.claude/skills/kazuo-create-plugin/references/plugin-examples.md`

## Files

- `manifest.json` — metadata, `defaultSettings`, `defaultWidgetConfig`
- `widget.html` / `js/widget.js` / `css/widget.css` — main UI
  - has its own "⚙" button + settings overlay built into widget.html/widget.js (host no longer provides a per-widget gear icon — see references/plugin-runtime-api.md section 1b)
- `settings.html` / `js/settings.js` / `css/settings.css` — same fields as the widget overlay (intentionally duplicated), used by the Plugin Manager screen and the pop-out window
- `locales/en.json5` / `locales/vi.json5` — UI strings, `widget.*`/`settings.*` namespaces
- `main/index.ts` (+ compiled `main/index.js`) — scans `~/.claude/projects/**/*.jsonl` across Windows + WSL, exposes list + exclude IPC channels
- `main/code-review-ipc.ts` / `main/code-review.ts` / `main/discovery.ts` — separate **Code & Plan
  Review** feature: main-process-owned review lifecycle, read-only source review store (Markdown
  reports, freshness fingerprints, handoff claims), and plan-file discovery/fingerprinting. Requires
  an ai-chat build that exposes `capabilities.readOnlyReview`/`startExecution` on
  `globalThis.__kazuoAiChat` (older ai-chat versions make `start-code-review` reject with a
  restart-required error, no unrestricted fallback). Details: `.claude/features/agent-review-plugin.md`
  (section "Code & Plan Review"), `.claude/features/code-plan-review-ipc.md`.
- `js/code-review.js` / `css/code-review.css` — standalone dialog UI for the Code & Plan Review
  feature, added alongside `widget.html`'s existing toolbar; does not modify `js/widget.js`. Details:
  `.claude/features/code-plan-review-ui.md`.

## Default settings

- `excludedWorkspaces` — default: `[]`
- `excludedSessions` — default: `[]`
- `scanWsl` — default: `true` (also scan WSL Claude config dirs over UNC)

## Task Manager link

Each session's "+ Task" button calls `invoke('plugin:task-manager:upsert', ...)`. Task Manager must be
installed + enabled for that button to work; otherwise Agent Review runs standalone. See
`.claude/features/agent-review-plugin.md`.

## Test

Open Kazuo AI → Plugin Manager — this plugin appears automatically (the host scans `plugins-library/` on its own, no registration needed) → Install → Enable.

After editing `main/index.ts` or `main/code-review*.ts`/`main/discovery.ts`, recompile before testing:
`npx tsc -p tsconfig.plugins.json` (run from the project root). Code & Plan Review's Electron
end-to-end behavior has not been re-verified live as of this doc update — see
`.claude/features/agent-review-plugin.md`'s "Code & Plan Review" section for what is and isn't
confirmed yet.