How nikhiltiwari1234/live-ai-assistant works
This small Python script repeatedly captures the screen, converts it to PNG bytes, and sends changed screenshots to Gemini for application/activity/error analysis. It prints the response and uses the prior screenshot and response to avoid unchanged-image requests and ask Gemini to suppress similar answers. The sampled source shows this workflow in one file; it does not show user-facing support actions beyond printed advice.
Main entry points: live_assistant.py Monitoring loop
Components 6
Other components
- Desktop
- Monitoring loop
live_assistant.py - Screen capture
live_assistant.py - PNG change check
live_assistant.py - Gemini API
- Console output
Connections 6
- Monitoring loop → Screen capture: captures
live_assistant.py - Screen capture → Desktop: reads screen
live_assistant.py - Monitoring loop → PNG change check: checks changes
live_assistant.py - Monitoring loop → Gemini API: submits image
live_assistant.py - Gemini API → Monitoring loop: returns analysis
live_assistant.py - Monitoring loop → Console output: prints response
live_assistant.py