langchain-aiOfficial·14 Agent Skills Included

skills-benchmarks

Benchmarks measuring how skill documentation improves coding accuracy

Measures how skill documentation design affects coding agent performance on LangChain, LangGraph, Deep Agents, and LangSmith tasks. Runs sandboxed Docker test cases covering agent building, tracing, evaluators, persistence fixes, and framework selection. Compares skill configurations against baselines with pass rates, turn counts, and LangSmith experiment tracking.

113|9

↗

npx skills add langchain-ai/skills-benchmarks --all -g -y

Available:

Directs the agent to invoke the correct LangChain, LangGraph, Deep Agents, or LangSmith skill before writing any code, and provides a build-trace-dataset-evaluate debugging workflow.

All Skills in This Repository (14)

Pure Emerald Level Indicators

📦 In Repo

langchain-ai

docker-patterns

Refactor Dockerfiles into multi-stage, non-root, scanned production images.

react-components

Generate scalable React UI components with TypeScript and hooks.

database-migrations

Apply versioned Up/Down migration scripts with rollback and validation support.

api-docs

Document RESTful APIs using OpenAPI 3.0+ templates with error schemas and versioning.

testing-patterns

Standardize Python unit and integration tests using Arrange-Act-Assert structure.

langchain-agents

Automate construction of LangChain agents using LangGraph patterns and context management.

langchain-oss-primer

Guide engineers through framework and agent archetype selection for LangChain projects.

langgraph-human-in-the-loop

Implement human-in-the-loop patterns in LangGraph graphs with interrupts and resume.

langgraph-persistence

Persist LangGraph state across conversations and subgraphs with configurable backends.

langchain-middleware

Coordinate human-in-the-loop approval and custom middleware for LangChain agents.

deep-agents-memory

Configure pluggable memory backends for ephemeral and persistent storage in Deep Agents.

langchain-rag

Orchestrate end-to-end RAG pipelines with LangChain loaders, splitters, embeddings, and vector stores.

Official

Advanced

Frequently Asked Questions

FAQPage Schema

How to install skills-benchmarks?▼

Run `npx skills add langchain-ai/skills-benchmarks --all -g -y` in your terminal to install all skills in this suite globally.

What do these benchmarks measure?▼

They measure how skill documentation design affects a coding agent's adherence to recommended LangChain, LangGraph, and LangSmith patterns, comparing pass rates across skill configurations.

How do I run a benchmark task?▼

Use pytest with a task and treatment flag, for example `uv run pytest tests/tasks/test_tasks.py --task=lc-basic --treatment=ALL_MAIN_SKILLS -v`. Docker and API keys for OpenAI, Anthropic, and LangSmith are required.

Which skills are included in this suite?▼

It includes production skills for LangChain fundamentals, RAG, middleware, dependencies, LangGraph persistence and human-in-the-loop, Deep Agents core, memory, and orchestration, plus LangSmith tracing, datasets, and evaluators.

Do the skills work in both Python and TypeScript?▼

Yes. Every skill provides equivalent Python and TypeScript implementations with matching commands, output formats, and error handling.