finds.dev← search

// the find

JoshuaC215/agent-service-toolkit

★ 4,350 · Python · MIT · updated Jun 2026

Full toolkit for running an AI agent service built with LangGraph, FastAPI and Streamlit

A batteries-included template for wrapping a LangGraph agent behind a FastAPI service with a Streamlit chat UI. It's for developers who want to ship a working agent endpoint fast without stitching together the FastAPI/LangGraph/streaming boilerplate themselves. Think of it as a reference architecture you clone and modify, not a library you import.

The streaming implementation handles both token-level and message-level streaming in a single client, which is non-trivial to get right. The multi-agent support via URL path routing is clean — add an agent to the dict, get an endpoint for free. Test coverage is genuine: unit, integration, and Docker e2e tests with codecov tracking, not just a few smoke tests. LangGraph Studio integration works out of the box with `langgraph dev`, which shortens the feedback loop when building graph logic.

The architecture is opinionated toward a single-process monolith: the FastAPI service and all agents run together, so a slow agent blocks others unless you're careful with async. Memory backends (Postgres, SQLite, MongoDB) are implemented as separate files but there's no abstraction layer — swapping backends means editing agent code, not just config. The Streamlit UI is convenient for demos but not something you'd ship to real users; the path to a production frontend is 'build it yourself'. The ChromaDB RAG agent is in the repo but its setup is in a separate doc with extra scripts, which means it's not actually plugged in by default — it's more of an example than a feature.

View on GitHub → Homepage ↗

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →