flynn/docs/plans/2026-02-16-voice-wake-and-capture-checklist.md

Voice Wake + Capture Checklist

Date: 2026-02-16 Status: completed

Scope

• Implement a practical Voice Wake / Talk Mode slice.
• Implement Camera / Screen Capture via host tools.

Completed

• Added audio.talk_mode config schema:
  • enabled (default false)
  • wake_phrase (default hey flynn)
  • timeout_ms (default 120000)
  • allow_manual_toggle (default true)
• Wired talk mode gating in src/daemon/routing.ts:
  • wake phrase activation with timed listen window
  • manual controls: /talk on, /talk off, /talk status
  • idle mode drops non-wake text when talk mode is enabled
• Added built-in capture tools in src/tools/builtin/capture.ts:
  • screen.capture
  • camera.capture
  • returns base64 image payloads (image/png or image/jpeg)
• Registered capture tools in builtin index and tool policy runtime/coding sets.
• Updated docs:
  • README.md
  • config/default.yaml
  • gap/roadmap docs in docs/plans/

Tests

• src/daemon/routing.test.ts (wake phrase/talk mode behavior)
• src/tools/builtin/capture.test.ts (capture tool command wrapping behavior)
• src/config/schema.test.ts (talk mode schema defaults/overrides)

Verification

• pnpm test:run src/daemon/routing.test.ts src/tools/builtin/capture.test.ts src/config/schema.test.ts src/tools/policy.test.ts
• pnpm typecheck