How sim-security/ai-teaching-agent-team works
The product is a Streamlit app that turns a learner’s topic into a learning package: a supervisor coordinates knowledge-base, roadmap, resource-curation, and practice-material stages, then presents the outputs and document links. LangGraph carries shared state between stages; agents use LLMs, and selected agents use search and Google Docs tools. The README documents tracing and model flexibility, but the excerpts do not show their full runtime wiring.
Main entry points: supervisor.py Supervisor Router · graph.py LangGraph Workflow · state.py Shared Teaching State · research_librarian.py Resource Guide
Components 17
Learner Interface
- Streamlit App
app.py
Agent Workflow
- LangGraph Workflow
src/graph.py - Shared Teaching State
src/state.py - Supervisor Router
src/supervisor.py - Knowledge Base
src/agents/professor.py - Learning Roadmap
src/agents/academic_advisor.py - Resource Guide
src/agents/research_librarian.py - Practice Materials
src/agents/teaching_assistant.py - Agent Tool Utilities
src/agents/utils.py
Tools and Integrations
- Google Docs Tools
src/tools/google_docs.py - Web Search Tool
src/tools/search.py
External Services
- Composio MCP
- Google Docs
- Web Search Provider
- LLM Provider
- LangSmith
Other components
- Learner
Connections 38
- Learner → Streamlit App: enters topic
README.md - Streamlit App → LangGraph Workflow: invokes workflow
app.py - Streamlit App → Google Docs Tools: loads tools
app.py - Streamlit App → Web Search Tool: loads tool
app.py - LangGraph Workflow → Shared Teaching State: uses schema
src/graph.py - LangGraph Workflow → Supervisor Router: registers node
src/graph.py - LangGraph Workflow → Knowledge Base: registers node
src/graph.py - LangGraph Workflow → Learning Roadmap: registers node
src/graph.py - LangGraph Workflow → Resource Guide: registers node
src/graph.py - LangGraph Workflow → Practice Materials: registers node
src/graph.py - Supervisor Router → Knowledge Base: routes first
README.md - Supervisor Router → Learning Roadmap: routes next
README.md - Supervisor Router → Resource Guide: routes next
README.md - Supervisor Router → Practice Materials: routes next
README.md - Knowledge Base → Supervisor Router: returns control
- Learning Roadmap → Supervisor Router: returns control
- Resource Guide → Supervisor Router: returns control
- Practice Materials → Supervisor Router: returns control
- Knowledge Base → Shared Teaching State: updates output
src/agents/professor.py - Learning Roadmap → Shared Teaching State: updates output
src/agents/academic_advisor.py - Resource Guide → Shared Teaching State: updates output
src/agents/research_librarian.py - Practice Materials → Shared Teaching State: updates output
src/agents/teaching_assistant.py - Shared Teaching State → Learning Roadmap: provides knowledge base
src/agents/academic_advisor.py - Shared Teaching State → Resource Guide: provides roadmap
src/agents/research_librarian.py - Shared Teaching State → Practice Materials: provides prior work
src/agents/teaching_assistant.py - Knowledge Base → Agent Tool Utilities: saves content
src/agents/professor.py - Learning Roadmap → Agent Tool Utilities: saves content
src/agents/academic_advisor.py - Resource Guide → Agent Tool Utilities: uses tools
src/agents/research_librarian.py - Practice Materials → Agent Tool Utilities: uses tools
src/agents/teaching_assistant.py - Resource Guide → Web Search Tool: searches resources
src/agents/research_librarian.py - Practice Materials → Web Search Tool: searches examples
src/agents/teaching_assistant.py - Agent Tool Utilities → Google Docs Tools: invokes tools
src/agents/utils.py - Google Docs Tools → Composio MCP: connects via MCP
src/tools/google_docs.py - Composio MCP → Google Docs: accesses documents
README.md - Web Search Tool → Web Search Provider: queries provider
src/tools/search.py - Streamlit App → LLM Provider: selects model
README.md - Streamlit App → LangSmith: configures tracing
app.py - LangGraph Workflow → LLM Provider: passes model to agents
src/graph.py