A modular multi-agent AI assistant for mobile applications, combining conversational intelligence, personalized recommendations, and behavior-driven user modeling in a unified frontend-backend architecture.
Each stack should be documented and runnable independently. Use the language-specific guides in docs/services/ to start and validate each runtime.
Current repository quality pattern:
- Keep API contract boundaries explicit.
- Keep source and tests paired by language conventions.
- Run lint, type checks, tests, and coverage before push.
Frontend steering highlights:
- New chat creation is handled from the sidebar Current Session action.
- Voice input triggers a dedicated speech service from the microphone button and streams text back live.
- Profile includes a help/forum section for feature lookup.
Local voice stack launch:
npm run dev
Manual fallback (two terminals):
npm --prefix backend run devnpm --prefix frontend run dev
The main branch is protected by required CI checks:
- Lint Gate
- Bandit Gate
- Test And Build Gate
For local enforcement before commit, install hooks once:
python -m pip install pre-commitpre-commit install
