LangChain has released its first set of skills, designed to provide AI coding agents with greater expertise in the open source LangChain ecosystem. These skills include building agents with LangChain, LangGraph, and Deep Agents. Internal tests show an increase in Claude Code's performance from 29% to 95% in performing specific tasks.

What are Skills?

Skills are instructions, scripts, and resources that improve the performance of coding agents in specialized domains. They are dynamically loaded through a 'progressive disclosure' mechanism: the agent retrieves a skill only when it is relevant to the task at hand. This approach contrasts with the practice of providing too many tools to an agent, which historically led to a drop in performance.

Skills are portable and shareable, consisting of markdown files and scripts that can be retrieved on demand. LangChain shares a set of skills that can be transferred to any coding agent that supports this functionality.

LangChain Skills

Within the langchain-skills repository, LangChain maintains a set of 11 skills, divided into three main categories:

  • LangChain: Guidance on how to use LangChain's create_agent(), middleware, and tool patterns. Fundamentals for working with the classic tool calling agent loop.
  • LangGraph: Guidance on how to use LangGraph's primitives and the benefits of its native support for Human In the Loop, durable execution, and more.
  • DeepAgents: Guidance on using the Deep Agents open source package and using its pre-built middleware and FileSystem.

Skill Impacts

Using skills led to significant improvements in Claude Code's performance on basic tasks related to LangChain, LangGraph, and DeepAgent.

Test Model Pass Rate
Claude Code without Skills Sonnet 4.6 25%
Claude Code with Skills Sonnet 4.6 95%

Pass rate was calculated using LangSmith evaluations. LangChain plans to open source the testing benchmark used.

Installation

To install these skills, you can use npx skills:

Local (current project):

npx skills add langchain-ai/langchain-skills --skill '*' --yes

Global (all projects):

npx skills add langchain-ai/langchain-skills --skill '*' --yes --global

To link skills to a specific agent (e.g. Claude Code):

npx skills add langchain-ai/langchain-skills --agent claude-code --skill '*' --yes --global

Conclusion

LangChain invites the community to use LangChain and LangSmith to improve the development experience with its ecosystem. The company plans to continue adding skills content as new features are added to its Open Source and LangSmith. In addition to these skills for LangChain open source, a set of LangSmith skills has also been released.