This toolkit provides a comprehensive solution for building and deploying AI agent services, targeting developers and researchers who want to leverage LangGraph for complex agentic workflows. It simplifies the process from agent definition to a user-facing Streamlit application, offering a robust template for rapid development and deployment.

How It Works

The project utilizes LangGraph for agent orchestration, enabling advanced features like human-in-the-loop interactions (interrupt()) and flow control (Command, langgraph-supervisor). A FastAPI service exposes the agent with both streaming and non-streaming endpoints, featuring a novel approach to handle token-based and message-based streaming. Pydantic is used for data structures and settings, ensuring type safety and clear configuration.

Quick Start & Requirements

• Install: pip install uv then uv sync --frozen (creates .venv), source .venv/bin/activate.
• Prerequisites: At least one LLM API key (e.g., OPENAI_API_KEY) set in a .env file.
• Run:
  • Service: python src/run_service.py
  • UI: streamlit run src/streamlit_app.py
• Docker: docker compose watch (recommended for development).
• Docs: LangGraph, FastAPI, Streamlit

Highlighted Details

• LangGraph v0.3 features including interrupt(), Command, and langgraph-supervisor.
• Advanced streaming support for both token-based and message-based outputs.
• Multiple agent support with distinct URL paths.
• Content moderation via LlamaGuard (requires Groq API key).
• Integrated LangSmith feedback mechanism.
• Comprehensive unit and integration tests.

Maintenance & Community

The project is actively maintained by JoshuaC215. Contributions are welcome via pull requests. Testing instructions are provided for local development.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive MIT license allows for commercial use and integration with closed-source applications.

Limitations & Caveats

Ollama support is experimental. Running tests requires the local development setup without Docker.