How sufian-01/multi-agent-scheduling-assistant works
This is a multi-agent scheduling and workplace-assistant application: users can manage calendar events, handle email, ask questions against a company FAQ, or get general conversational help. The LangGraph workflow classifies a request, routes it to a specialist, and lets tool-using specialists dispatch tools before returning a response. The sampled sources show those routes and the calendar, Gmail, and RAG tool implementations; the README is empty, and the frontend and app entry-point internals were not excerpted, so their connections are shown only as unverified.
Main entry points: graph.py Workflow graph · booking_agent.py Booking agent · email_agent.py Email agent · rag_agent.py Knowledge agent
Components 19
User interface
- Frontend
frontend/main.py - Application entry
app.py
Agent orchestration
- Workflow graph
graph.py - Intent triage
agent/triage_agent.py
Assistant capabilities
- Booking agent
agent/booking_agent.py - Email agent
agent/email_agent.py - Knowledge agent
agent/rag_agent.py - General agent
agent/general_agent.py
Tools and integrations
- Calendar tools
tools/calendar_tools.py - Gmail tools
tools/gmail_tools.py - FAQ retrieval
tools/rag_tools.py - Company FAQ PDF
knowledge_base/company_faq.pdf - Google auth
tools/google_auth.py
Shared runtime
- Agent prompts
prompts.py - Conversation state
state.py - Runtime config
config.py
Other components
- User
- Google APIs
- Gemini model
Connections 37
- User → Frontend: submits request
- Frontend → Application entry: sends request
- Application entry → Workflow graph: invokes workflow
app.py - Workflow graph → Intent triage: starts at
graph.py - Workflow graph → Booking agent: routes intent
graph.py - Workflow graph → Email agent: routes intent
graph.py - Workflow graph → Knowledge agent: routes intent
graph.py - Workflow graph → General agent: routes intent
graph.py - Intent triage → Gemini model: classifies with
agent/triage_agent.py - Booking agent → Calendar tools: binds tools
agent/booking_agent.py - Email agent → Gmail tools: binds tools
agent/email_agent.py - Knowledge agent → FAQ retrieval: binds tool
agent/rag_agent.py - Workflow graph → Calendar tools: registers tools
graph.py - Workflow graph → Gmail tools: registers tools
graph.py - Workflow graph → FAQ retrieval: registers tool
graph.py - Calendar tools → Google APIs: uses Calendar API
tools/calendar_tools.py - Gmail tools → Google APIs: uses Gmail API
tools/gmail_tools.py - Calendar tools → Google auth: gets credentials
tools/calendar_tools.py - Gmail tools → Google auth: gets credentials
tools/gmail_tools.py - Google auth → Google APIs: authenticates
tools/google_auth.py - FAQ retrieval → Company FAQ PDF: loads PDF
tools/rag_tools.py - Knowledge agent → Gemini model: invokes model
agent/rag_agent.py - Booking agent → Gemini model: invokes model
agent/booking_agent.py - Email agent → Gemini model: invokes model
agent/email_agent.py - Intent triage → Agent prompts: uses prompt
agent/triage_agent.py - Booking agent → Agent prompts: uses prompt
agent/booking_agent.py - Email agent → Agent prompts: uses prompt
agent/email_agent.py - Knowledge agent → Agent prompts: uses prompt
agent/rag_agent.py - Intent triage → Conversation state: reads and updates
agent/triage_agent.py - Booking agent → Conversation state: reads and updates
agent/booking_agent.py - Email agent → Conversation state: reads and updates
agent/email_agent.py - Knowledge agent → Conversation state: reads and updates
agent/rag_agent.py - Workflow graph → Conversation state: builds over
graph.py - Intent triage → Runtime config: loads API key
agent/triage_agent.py - Booking agent → Runtime config: loads API key
agent/booking_agent.py - Email agent → Runtime config: loads API key
agent/email_agent.py - Knowledge agent → Runtime config: loads API key
agent/rag_agent.py