Deep Agents Core Skill: Build a LangChain Deep Agent with Claude

Teaches Claude how to build a LangChain Deep Agent: the agent setup call, built-in middleware, skills and storage backends, in Python or TypeScript.

Installs: 16.3K; on skills.sh
Stars: 1.3K; on GitHub
Updated: Oct 2026; last repo commit

Deep Agents Core is a LangChain guide that Claude reads when you build any Deep Agents application. It explains what the harness gives you by default (planning, file tools, subagents) and shows Python and TypeScript code for creating an agent, loading skills and setting up memory.

It is a reference and bundles only SKILL.md, no scripts. You still need the deepagents package and a model provider key in your own project.

Good for

Developers starting a first Deep Agents project
Teams choosing between Deep Agents and LangChain create_agent
Anyone adding SKILL.md skills or long-term memory to a Deep Agent

Install deep-agents-core

Claude Code

Copy this prompt.

Install the deep-agents-core skill in Claude Code. Run this command:

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

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.

deep-agents-core stands alone: its SKILL.md does not call other skills. The plugin route installs all 21 skills in langchain-skills (plugin.json points at config/skills/ and declares no hooks or MCP servers). The README marks the project as early development. Pick one route, not both.

From the project README. Source.

What's in the skill

When it loads

INVOKE THIS SKILL when building ANY Deep Agents application. Covers create_deep_agent(), harness architecture, SKILL.md format, and configuration options.

Sections in SKILL.md

SKILL.md Format
Directory Structure
SKILL.md Format
Overview
When to Use
Instructions
What Agents CAN Configure
What Agents CANNOT Configure

Key facts

Author: LangChain
License: MIT
Installs on skills.sh: 16,280 (Nov 2, 2026)
GitHub stars: 1,320 for the whole langchain-ai/langchain-skills repo (Nov 2, 2026)
Last commit: Oct 29, 2026
What is inside: 1 file: SKILL.md. No scripts or other files.
Starts: Claude loads it when a task matches its description, which begins "INVOKE THIS SKILL"
Repo status: The README marks the project as early development; APIs and skill content may change

What it includes

Compares Deep Agents with LangChain create_agent
Use Deep Agents for multi-step planning, large context, subagents and memory across sessions. Use create_agent for simple single-purpose tasks.
When to use Deep Agents
Six middleware: todo list, filesystem, subagents, approval, skills, memory
A table of which middleware to use for which need. Approval needs a checkpointer and memory needs a Store.
Middleware selection
write_todos, file tools (ls, read_file, write_file, edit_file, glob, grep) and task
Every deep agent gets planning, filesystem and delegation tools without extra setup.
Built-in tools
create_deep_agent with a custom tool, then with all options
Python and TypeScript examples for a basic agent and a fully configured one.
Try: “Create a deep agent with a weather tool and a thread_id based conversation.”
Basic and full configuration examples
Folder layout, front matter and progressive loading for agent skills
Explains how skills load on demand, and when to use skills versus AGENTS.md memory.
SKILL.md format
Load skills from the filesystem or from a Store
Two examples: skills on disk, and skills loaded into a Store.
Skills backends
What you cannot change, plus checkpointer, store, thread_id and skill-description fixes
You cannot remove the core middleware or rename its tools. Fixes cover missing checkpointers, missing stores, no thread_id, missing front matter and subagent skills.
Boundaries and fixes

From the official docs

Questions

What does the deep-agents-core skill do?

It gives Claude LangChain's guidance for building a Deep Agents app: how create_deep_agent works, what the harness adds, how skills load and how to avoid common setup mistakes.

How do I install it?

Use the install box at the top of this page. The langchain-skills README also documents a Claude Code plugin and an install script for Claude Code and the Deep Agents CLI.

Does it need an API key?

The skill itself needs none. The agents you build need a model provider key in your own project; the README says to export OPENAI_API_KEY or ANTHROPIC_API_KEY.

Is it safe to install?

It is a single text file with examples and no scripts, so it only adds instructions for Claude. Review the code Claude writes from it before running it.

Is this the same as using SKILL.md files in Claude Code?

No. Part of this skill explains the SKILL.md format for Deep Agents, which loads skills through its own middleware. It is about building agents, not about how Claude Code loads this page's skill.

What should I install with it?

The README lists deep-agents-memory and deep-agents-orchestration as the other Deep Agents skills, and ecosystem-primer as a start-here guide. SKILL.md does not require any of them.

Information

updated Nov 2, 2026

Repository: langchain-ai/langchain-skills
Folder: config/skills/deep-agents-core

langchain-ai/langchain-skills skills.sh

Sources

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

Show all 3

github.com · github.com
github.com · github.com
github.com · github.com