How yobo-dance/wisdom-cowork-agent works
Wisdom Cowork Agent is an unfinished desktop-assistance project; the README describes helping with daily desktop tasks but does not specify a complete end-to-end workflow. The clearest implemented path is push-to-talk speech capture and transcription in src/core/STT.py. The repository also has speech output, a brain module, and tool modules for apps, browsing, files, and system control, but their implementations or connections were not shown. Dashed edges mark this unverified intended integration; the learning exercises are excluded.
Main entry points: brain.py Agent brain · STT.py Speech recognition · main.py Application entry · TTS.py Speech output
Components 11
Agent runtime
- Application entry
src/main.py - Agent brain
src/core/brain.py
Voice I/O
- Speech recognition
src/core/STT.py - Speech output
src/core/TTS.py
Desktop tools
- App control
src/tools/app_control.py - Browser tools
src/tools/browser.py - File management
src/tools/file_manager.py - System control
src/tools/system_control.py
Other components
- Desktop user
- Audio input
- Whisper model
Connections 10
- Desktop user → Speech recognition: holds hotkey
src/core/STT.py - Speech recognition → Audio input: captures audio
src/core/STT.py - Speech recognition → Whisper model: transcribes audio
src/core/STT.py - Application entry → Agent brain: intended dispatch
- Agent brain → Speech recognition: intended speech input
- Agent brain → Speech output: intended speech output
- Agent brain → App control: intended tool use
- Agent brain → Browser tools: intended tool use
- Agent brain → File management: intended tool use
- Agent brain → System control: intended tool use