All tools
Automation & agentsUpdated today

LangChain

By LangChain

MIT-licensed Python framework, around 148,000 GitHub stars and roughly 37 million weekly PyPI downloads, for building agents and LLM applications through a single create_agent harness and one model interface across OpenAI, Anthropic, Google, Bedrock, Ollama, and other providers, with middleware for guardrails, retries, and tool policies; agents run on LangGraph for persistence and human-in-the-loop steps, and a JavaScript version ships as LangChain.js.

Best for

  • switching model providers without rewriting agent code
  • adding guardrails, retries, and tool policies as middleware
  • connecting an agent to vector stores, retrievers, and third-party tools
  • agents that pause for human approval and resume later
  • a starting point that grows into LangGraph orchestration
Recent changes
  • Oct 28, 2026Added to the automation category: MIT-licensed agent framework, roughly 148,000 GitHub stars and 37 million weekly PyPI downloads, built around create_agent, a provider-neutral model interface, and middleware, running on LangGraph. Its observability product, LangSmith, was already in the catalogue under evals; the framework itself was not.

Other Automation & agents