LangGraph
Low-level Python framework for stateful, long-running agents
What it is
A Python orchestration library for building agents as graphs of steps with persisted state. It provides durable execution, checkpointing and human-in-the-loop hooks rather than ready-made agents; a JS/TS sibling exists in a separate repo. For engineers who want fine control over agent control flow.
Good for
- Build an agent that survives restarts
- Add human approval steps to a workflow
- Model multi-step agent logic as a graph
Install or run
pip install -U langgraphCommand as given in the project README. Read the repository before you run anything on your machine.
What you need
Python; an LLM provider API key
Sources
- https://github.com/langchain-ai/langgraph (project README, read 10 Nov 2026)
- https://api.github.com/repos/langchain-ai/langgraph (GitHub API, read 13 Nov 2026)
The summary is ours, written from the README. Spotted an error? See how we work.