langchain-skills

Why we included this project

Maintained by the LangChain team itself, this pack of agent skills is for people who build agents with LangChain, LangGraph, or Deep Agents and want their coding assistant to write code that fits those frameworks. The skills install via npx or as a Claude Code plugin and work in Claude Code, Cursor, Windsurf, and other tools that support the Agent Skills spec, covering the parts of the stack you actually touch: Python and TypeScript quickstarts, dependency and version management, RAG pipelines, LangGraph state and persistence, human-in-the-loop approval flows, and Deep Agents orchestration and memory. There's also an eval-engineering skill for building and running Harbor-based evaluations on an existing agent. Because the content tracks the official LangChain docs, it tends to stay aligned with current APIs, which is useful given how quickly those change. One caveat: the project is still in early development, so APIs and skill content may shift.