fastapi-langgraph-agent-production-ready-template

What it is

A production-ready template for building AI agent backends with FastAPI and LangGraph. It handles stateful conversations, long-term memory, tool calling, observability, rate limiting, and JWT authentication.

How it works

• Includes a LangGraph stateful agent with checkpointing, tool calling, and human-in-the-loop support.
• Long-term memory via mem0 and pgvector, with semantic search per user and cache backing.
• LLM service with circular model fallback, exponential backoff retries, and a total timeout budget.
• Observability through Langfuse tracing, Prometheus metrics, and Grafana dashboards.
• JWT authentication with session management and rate limiting via slowapi.
• Alembic migrations and optional Valkey/Redis cache layer.
• Structured logging with request, session, and user context on every line.

Getting started

• Quickstart commands from README:

git clone <repo-url> my-agent && cd my-agent
cp .env.example .env.development   # fill in your keys
make install
make docker-up                     # starts API + PostgreSQL

• Open API docs at: http://localhost:8000/docs

Recent releases

• Latest releases: none

Traction

• Stars: 2553
• Forks: 599
• Open issues: 10

License

• MIT

Configuration and usage notes

• LLM providers: OpenAI via LLMRegistry currently; multi-provider support planned via LangChain’s init_chat_model. Configure via DEFAULT_LLM_MODEL in .env.development.
• Atlas Cloud integration: drop-in LLM backend via OPENAI_BASE_URL and OPENAI_API_KEY; example values provided in README.
• Docker recommended but not required; use make docker-up for API + PostgreSQL.