langchain-skills

LangChain Skills

⚠️ — This project is in early development. APIs and skill content may change.

Agent skills for building agents with LangChain, LangGraph, and Deep Agents.

For LangSmith-specific trace and dataset workflows, use langsmith-skills.

Supported Coding Agents

These skills can be installed via npx skills for any agent that supports the Agent Skills specification, including Claude Code, Cursor, Windsurf, and more.

Installation

Quick Install

Using 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

---

Claude Code Plugin

Install directly as a Claude Code plugin:

/plugin marketplace add langchain-ai/langchain-skills
/plugin install langchain-skills@langchain-skills

---

Install Script (Claude Code & Deep Agents CLI only)

Alternatively, clone the repo and use the install script:

# Install for Claude Code in current directory (default)
./install.sh

# Install for Claude Code in a specific project directory
./install.sh ~/my-project

# Install for Claude Code globally
./install.sh --global

# Install for Deep Agents CLI in a specific project directory
./install.sh --deepagents ~/my-project

# Install for Deep Agents CLI globally (includes agent persona)
./install.sh --deepagents --global

Flag / Argument	Description
DIRECTORY	Target project directory (default: current directory, ignored with --global)
--claude	Install for Claude Code (default)
--deepagents	Install for Deep Agents CLI
--global, -g	Install globally instead of current directory
--force, -f	Overwrite skills with same names as this package
--yes, -y	Skip confirmation prompts

Usage

After installation, set your API keys:

export OPENAI_API_KEY=<your-key>      # For OpenAI models
export ANTHROPIC_API_KEY=<your-key>   # For Anthropic models

Then run your coding agent from the directory where you installed (for local installs) or from anywhere (for global installs).

Eval Engineering

To install only the eval-engineering skill:

npx skills add langchain-ai/langchain-skills --skill eval-engineering --yes

Or ask Codex: “Install the eval-engineering skill from langchain-ai/langchain-skills.”

Eval tasks require Harbor. Run Harbor locally with Docker or use a supported cloud environment.

Then ask your coding agent:

Use eval-engineering to inspect this agent and propose two or three Harbor evals. Traces for this agent can be found here [Tracing Project/Location].

Available Skills (15)

Getting Started

• ecosystem-primer - Start-here primer: framework selection (LangChain vs LangGraph vs Deep Agents), env setup, and which skill to load next
• langchain-dependencies - Full package version and dependency management reference (Python + TypeScript)

Deep Agents

• deep-agents-core - Agent architecture, harness setup, and SKILL.md format
• deep-agents-memory - Memory, persistence, filesystem middleware
• deep-agents-orchestration - Subagents, task planning, human-in-the-loop
• managed-deep-agents - Managed Deep Agents: deploy with the CLI, use the SDKs, stream runs, connect MCP tools, and build React useStream UIs

LangChain

• langchain-fundamentals - Agents with create_agent, tools, structured output, middleware basics
• langchain-middleware - Human-in-the-loop approval, custom middleware, Command resume patterns
• langchain-rag - RAG pipeline (document loaders, embeddings, vector stores)

LangGraph

• langgraph-fundamentals - StateGraph, nodes, edges, state reducers
• langgraph-persistence - Checkpointers, thread_id, cross-thread memory
• langgraph-cli - CLI lifecycle: scaffold, dev, build, deploy, langgraph.json config
• langgraph-human-in-the-loop - Interrupts, human review, approval workflows

Evaluation

• eval-engineering - Build, run, and audit Harbor evals for an existing agent with user approval

Utilities

• swarm - Dispatch independent work items in parallel and aggregate the results