LangChain
A framework for building agents and model-powered applications from interoperable components. It is a development library, not a ready-to-use autonomous assistant.
Utility: Frameworks; kind: framework.
Stars: 148k+; Starred in September: 2k+.
Metrics updated: 31 Oct 2026. Profile reviewed: 31 Oct 2026.
Utilities
- Connect models, tools, and retrieval components in application code.
- Build an agent with explicit provider and integration choices.
Who it is for
- Developers implementing their own agent applications.
Requirements
- Install the Python package and provider-specific integrations you need.
- Supply model credentials or a compatible local endpoint.
Limitations
- Application behavior depends on the chosen tools, prompts, and model.
- Advanced stateful orchestration is a separate design task; the README points to LangGraph.
Cost and privacy notes
- The MIT framework does not include model usage or managed observability/deployment.
- Tool inputs and traces can contain sensitive data; review all configured providers.
Illustrative example (untested)
Illustrative, not tested: build a small agent with a read-only catalog tool and validate its structured responses.
License
MIT; reviewed the repository LICENSE.