Repolode
Submit
langchain-ai

langchain-ai/langgraph

A low-level Python framework for building stateful, long-running agents with durable execution, human-in-the-loop and memory.

View on GitHubWebsite
Stars
43,568
Forks
7,439
Open issues
497
Last commit
7 hours ago

What is it?

LangGraph is a low-level orchestration framework for building, managing and deploying long-running, stateful agents and workflows. It provides the underlying infrastructure rather than a high-level agent abstraction, so developers control how state flows through each step of an agent.

Agents built with LangGraph can persist through failures and resume from where they stopped. Developers can pause execution to inspect or modify agent state, and the framework supports both short-term working memory within a run and long-term memory across sessions. Execution traces can be visualized through LangSmith.

LangGraph can be used on its own or alongside LangChain's integrations and components. Deep Agents, a higher-level package built on LangGraph, is available for agents that plan and use subagents. A JavaScript/TypeScript version, LangGraph.js, is also maintained separately.

Key features

  • 01Durable execution that resumes agents from where they left off after failures
  • 02Human-in-the-loop interrupts for inspecting and modifying agent state
  • 03Short-term and long-term memory for stateful agents
  • 04Graph-based workflow design with support for branching and subgraphs
  • 05Integration with LangSmith for tracing and debugging

Use cases

  • Building multi-step AI agents that must survive crashes or run for long periods
  • Adding human approval steps to agent actions before they execute
  • Creating multi-agent systems where agents coordinate through shared state

Quick start

  1. 1

    Install LangGraph from PyPI

    pip install -U langgraph
  2. 2

    Open the LangGraph Quickstart guide to build your first agent

Things to know

  • Low-level framework, so building agents requires writing more code and learning its graph concepts.
  • Some production features such as deployment are tied to the LangSmith platform.

Topics

#agents#llm#workflow-orchestration#stateful#python#multiagent#langchain#ai#ai-agents#chatgpt

FAQ

What is langgraph?
A low-level Python framework for building stateful, long-running agents with durable execution, human-in-the-loop and memory. LangGraph is a low-level orchestration framework for building, managing and deploying long-running, stateful agents and workflows. It provides the underlying infrastructure rather than a high-level agent abstraction, so developers control how state flows through each step of an agent.
Is langgraph free to use? What license does it have?
langgraph is open source under the MIT license. Check the license terms before using it in commercial products.
Is langgraph still maintained?
Actively maintained. The last commit to langgraph was on Oct 20, 2026.
What can I use langgraph for?
Building multi-step AI agents that must survive crashes or run for long periods Adding human approval steps to agent actions before they execute Creating multi-agent systems where agents coordinate through shared state
How do I install langgraph?
Install LangGraph from PyPI: pip install -U langgraph Open the LangGraph Quickstart guide to build your first agent
Where can I get langgraph?
The source code and installation instructions are on GitHub: https://github.com/langchain-ai/langgraph

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects