How univac-1/meeting-ai-agent works
Main Components
• Frontend Application:
• Located under the apps/frontend directory.
• Built using TypeScript/React, with pages clearly laid out (e.g., 404, home, join, meeting, meetingInfo, meetingSchedule).
• Organized as page components alongside shared resources (common, components, router, store).
• Cloud Run Service:
• Located under the apps/cloudrun directory.
• Contains Python-based modules that implement various agents (feedback_agent, intervention_check_agent, etc.), services, and utility functions.
• Modules are responsible for tasks such as handling meeting data, managing AI functionalities (e.g., feedback and intervention operations), and updating service states.
• Deployment Infrastructure:
• GitHub Actions workflows are defined (in .github/workflows) for deploying both the Cloud Run backend and Firebase Hosted frontend.
• Additional config files such as firebase.json and Dockerfiles in both frontend and backend indicate containerized deployment.
• External Services:
• Firebase Hosting is used to serve the frontend.
• Cloud Run provides the backend service.
Relationships and Interactions
• Frontend-to-Backend Communication:
• The frontend interfaces with the Cloud Run service likely via REST APIs or similar endpoints (as indicated by the organization of both applications).
• Internal Component Interactions within the Cloud Run Service:
• Several Python modules in the “agent” directory work together (e.g., feedback_agent, intervention_check_agent) to process meeting-related events.
• Deployment Pipeline:
• GitHub Actions pipeline triggering deployments based on tag patterns (“b-“ for Cloud Run and “f-“ for Firebase).
Architectural Patterns and Technologies
• Architectural Patterns:
• Separation of concerns: Frontend is modularized by pages, and the backend is structured into agents and service modules.
• Containerization: Dockerfiles in both deployments suggest that both environments are containerized.
• Serverless/Managed Infrastructure: Utilizing Cloud Run and Firebase Hosting implies a serverless or managed services approach.
• Technologies/Frameworks:
• Frontend: React (with TypeScript), state management (Redux or similar store pattern), routing, and a well-organized component structure.
• Backend: Python for Cloud Run services, employing modules for business logic and agent interactions.
• Deployment: GitHub Actions for CI/CD automation, Firebase Hosting & Cloud Run as deployment targets.