fastapi -langgraph-template

This MCP server template provides a FastAPI-based implementation designed for agentic orchestration with LangGraph and MCP. It exposes a REST API that adheres to the MCP protocol, enabling inspectors and clients to interact with the server to obtain context, manage state, and drive agent workflows. The integration with LangGraph supports customizable agent orchestration pipelines, streaming UX, and persisted chat history/state. To use it, start the server (see installation) and interact with the MCP endpoints via HTTP clients or the provided SDKs. You can also leverage LangGraph-backed agents to compose complex workflows, concatenate multiple context sources, and observe LLM metrics through LangFuse-compatible tooling if you enable those features in your environment.

Prerequisites:

• Python 3.9+ installed on your system
• Access to a terminal/command prompt
• Optional: Docker and Docker Compose if you prefer containerized runs

Installation steps:

1. Clone the repository or download the template code: git clone https://github.com/nicholasgoh/fastapi-mcp-langgraph-template.git cd fastapi-mcp-langgraph-template

2. Set up a Python virtual environment and install dependencies: python -m venv venv source venv/bin/activate # On Windows use: venv\Scripts\activate pip install -r requirements.txt

3. Configure environment variables (example):

   • Create a .env file or export variables in your shell export ENV=development export DATABASE_URL=postgresql://user:password@localhost:5432/dbname export LANGGRAPH_API_KEY=your-langgraph-api-key export MCP_API_BASE_URL=http://localhost:8000

4. Run the server (development mode): uvicorn fastapi_mcp_langgraph_template.main:app --reload --host 0.0.0.0 --port 8000

Optional containerized run with Docker Compose (if provided in the repo): docker compose -f docker-compose.yaml up --build

Notes:

• If you are using a database, ensure the database server is running and accessible via DATABASE_URL.
• The server exposes MCP endpoints; you can use MCP client libraries to interact according to the MCP protocol.
• For production deployments, consider configuring a reverse proxy (Nginx) and a process manager (gunicorn/uvicorn workers) as appropriate.

Tips and common issues:

• Ensure the Python environment matches the project’s required versions (Python 3.9+).
• If LangGraph integration is optional in your setup, you can disable or skip related features via environment configuration.
• When using Docker, map ports appropriately and set DATABASE_URL and API keys as secrets or environment variables in your orchestrator.
• If you encounter MCP protocol negotiation errors, verify that the Inspector and server versions are compatible and that the MCP handshake endpoints are reachable.
• Check logs for any database connection errors, as these are a common startup failure point when not properly configuring DATABASE_URL.
• Consider enabling reloads during development to reflect code changes quickly, and switch to a production ASGI server setup for deployments.