langchain-architecture skill (wshobson) for Claude Code: build LangGraph agents

langchain-architecture helps Claude design LLM apps and agents with LangChain 1.x and LangGraph.

Installs: 13.4K; on skills.sh
Stars: 42.1K; on GitHub
Files: 1; besides SKILL.md
Updated: Oct 2026; last repo commit

langchain-architecture is a skill from wshobson/agents. It loads when you build LangChain apps, AI agents or multi-step LLM workflows. SKILL.md lists the 1.x package layout, then covers LangGraph agents, state, memory, document processing and LangSmith tracing, with a ReAct agent quick start, testing examples and performance tips.

references/details.md adds four architecture patterns (RAG, structured tools, a StateGraph workflow and multi-agent orchestration), memory, callbacks and streaming. The examples are Python and will need LangChain packages and model or vector store keys in your own project. The skill bundles no scripts.

Good for

Python developers building agents with LangGraph
Teams adding RAG, memory or multi-agent workflows to an app
Developers who want tests and tracing for LLM apps

Install langchain-architecture

Claude Code

Copy this prompt.

Install the langchain-architecture skill in Claude Code. Run this command:

npx skills add wshobson/agents --skill langchain-architecture

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

Built from the package data we hold. Not yet tested by us. Source.

What's in the skill

When it loads

Design LLM applications using LangChain 1.x and LangGraph for agents, memory, and tool integration. Use when building LangChain applications, implementing AI agents, or creating complex LLM workflows.

Sections in SKILL.md

When to Use This Skill
Package Structure (LangChain 1.x)
Core Concepts
1. LangGraph Agents
2. State Management
3. Memory Systems
4. Document Processing
5. Callbacks & Tracing
Quick Start
Modern ReAct Agent with LangGraph
Detailed patterns and worked examples
Testing Strategies
Performance Optimization
1. Caching with Redis
2. Async Batch Processing
3. Connection Pooling

Key facts

Author: wshobson (GitHub account that owns the repo)
Source repo: wshobson/agents
Installs on skills.sh: 13,374 as of 2026-11-02
GitHub stars: 42,081 (whole repo) as of 2026-11-02
License: MIT
Last repo commit: 2026-10-30
Files: 2 files including SKILL.md (plus references/details.md)
skills.sh audits: Gen Agent Trust Hub pass, Socket pass, Snyk pass

What it includes

Lists the LangChain 1.x packages: core, community, LangGraph and provider packages.
Package structure
StateGraph, durable execution, human-in-the-loop, memory and checkpointing.
LangGraph agents
ReAct, plan-and-execute, multi-agent supervisor and tool-calling with Pydantic schemas.
Agent patterns
Buffer, summary, token-window and vector-store memory, plus LangGraph checkpointers.
Memory systems
Loaders, text splitters, vector stores and retrievers.
Document processing
LangSmith for request logging, token usage, latency and trace views.
Network · Callbacks and tracing
A Python agent with Pydantic-schema tools, a checkpointer and a thread ID.
Quick start: ReAct agent
pytest examples for tool selection and memory persistence.
Testing strategies
Redis LLM cache, async batch processing and a reused Pinecone client.
Network · Performance optimization
RAG, structured tools, StateGraph workflow, multi-agent, memory and streaming.
Detailed patterns (references/details.md)

From the official docs

Permissions

Low risk

The skill is Markdown only, but generated LLM apps use model, vector store and tracing keys. Keep keys in environment variables, and approve commands before they run.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

Tracing can send data to LangSmith
The details file shows enabling LangSmith tracing with LANGCHAIN_TRACING_V2 and an API key. Turn it on only if you accept traces leaving your app. github.com
Examples call model and vector store services
The quick start and performance examples use provider integrations and a Pinecone key from the environment. Those calls run only when you run the code. github.com
The plugin route adds more than this skill
The llm-application-dev plugin folder holds 8 skills, 3 agent file(s) and 3 command file(s), plus plugin.json files. The repo tree shows no hooks or MCP config in it. The skills CLI route installs only this skill. github.com

Questions

How do I install the langchain-architecture skill?

Use the install box above. The skills CLI command installs only this skill. The plugin route installs llm-application-dev, which also holds seven other skills, three agent files and three command files.

Does langchain-architecture need an API key?

The skill does not ask for one. The code it helps you write uses model and vector store keys, such as a Pinecone key read from an environment variable, and optionally a LangSmith key for tracing.

Does it send my data to LangSmith?

The details file shows LangSmith tracing switched on with LANGCHAIN_TRACING_V2 and an API key, after which LangChain and LangGraph operations are traced automatically. Whether that runs depends on the code and environment variables in your own project.

Which LangChain version does it target?

LangChain 1.x with LangGraph. SKILL.md lists the package structure for the 1.x line and calls LangGraph the standard way to build agents.

Does it work in the Claude app or Codex?

The install box lists a zip upload for the Claude app and a plugin route for Codex. The wshobson/agents README names Claude Code, Codex CLI, Cursor, OpenCode, Antigravity CLI, GitHub Copilot and Pi as supported tools.

Is langchain-architecture safe to install?

The skill is Markdown files with no scripts, so installing it alone adds instructions only. The plugin route adds more skills, agents and commands, listed in the safety section. Keys in generated code are your responsibility, so keep them in environment variables.

Information

updated Nov 2, 2026

Repository: wshobson/agents
Folder: plugins/llm-application-dev/skills/langchain-architecture

wshobson/agents skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 6

wshobson/agents README · github.com
Claude Code skills · code.claude.com
Using skills in Claude · support.claude.com
SKILL.md · github.com
details.md · github.com
langchain-architecture on skills.sh · skills.sh