How yobo-dance/project-google-hackathon works
Sports Concierge generates personalized training plans through a security review and approval gate. The primary documented flow is the Streamlit dashboard: collect athlete details and context, gather email and drive metadata, scan files, assess risk, pause for human approval when needed, then generate and display a plan with an audit trail. The repository also contains a separate ADK/FastAPI serving surface with A2A and reasoning-engine routes. Its connections are limited to relationships shown in excerpts or the SOURCE INDEX; the README describes some dashboard capabilities whose implementation was not sampled.
Main entry points: app.py Streamlit dashboard · fast_api_app.py FastAPI app · guard.py Guard agent · reasoning_engine_adapter.py Reasoning routes
Components 20
Dashboard workflow
- Streamlit dashboard
app.py - Training plan
Agent workflow
- Guard agent
agents/guard.py - Coach agent
agents/coach.py
Context and security tools
- Email context
tools/server_mcp.py - Drive files
tools/server_mcp.py - MCP scan tool
tools/server_mcp.py - Security scanner
security/sandbox.py
Audit persistence
- Audit trail
database/manager.py
ADK serving
- FastAPI app
sports-concierge/app/fast_api_app.py - ADK agent
sports-concierge/app/agent.py - A2A routes
sports-concierge/app/app_utils/a2a.py - Reasoning routes
sports-concierge/app/app_utils/reasoning_engine_adapter.py - Shared services
sports-concierge/app/app_utils/services.py - Telemetry
sports-concierge/app/app_utils/telemetry.py - Feedback schema
sports-concierge/app/app_utils/typing.py
Other components
- Athlete
- Plan model
- Supabase
- ADK client
Connections 23
- Athlete → Streamlit dashboard: enters details
README.md - Streamlit dashboard → Guard agent: requests assessment
app.py - Guard agent → Email context: gathers context
README.md - Guard agent → Drive files: gathers context
README.md - Guard agent → MCP scan tool: requests scans
agents/guard.py - MCP scan tool → Security scanner: delegates scan
tools/server_mcp.py - Guard agent → Streamlit dashboard: returns risk
README.md - Athlete → Streamlit dashboard: approves or denies
README.md - Streamlit dashboard → Coach agent: requests plan
README.md - Coach agent → Plan model: generates with
README.md - Plan model → Streamlit dashboard: returns plan
README.md - Streamlit dashboard → Training plan: displays and downloads
README.md - Streamlit dashboard → Audit trail: logs events
app.py - Audit trail → Supabase: writes events
database/manager.py - ADK client → FastAPI app: calls API
- FastAPI app → ADK agent: loads agent
sports-concierge/app/fast_api_app.py - FastAPI app → A2A routes: attaches routes
sports-concierge/app/fast_api_app.py - FastAPI app → Reasoning routes: attaches routes
sports-concierge/app/fast_api_app.py - FastAPI app → Shared services: uses services
sports-concierge/app/fast_api_app.py - FastAPI app → Telemetry: sets up
sports-concierge/app/fast_api_app.py - FastAPI app → Feedback schema: validates feedback
sports-concierge/app/fast_api_app.py - Reasoning routes → ADK agent: dispatches methods
sports-concierge/app/app_utils/reasoning_engine_adapter.py - Reasoning routes → Shared services: reuses services
sports-concierge/app/app_utils/reasoning_engine_adapter.py