ecosystem-primer Skill (LangChain) for Claude Code: What It Does and How to Install It

Makes your coding agent choose between LangChain, LangGraph and Deep Agents before it writes any agent code.

Installs: 8.4K; on skills.sh
Stars: 1.3K; on GitHub
Updated: Oct 2026; last repo commit

This is LangChain's own start-here skill, from the vendor's own repo. It gives a four-step decision table for picking a layer, profiles for each tool, the LangSmith environment variables to set and where the docs live. Then it tells the agent which skill from the same repo to load next.

It triggers for any LangChain, LangGraph or Deep Agents project, before other skills. It bundles no scripts. The skill says the primer alone does not carry the current API, so the layer-specific skills from the same repo are part of the setup.

Good for

Developers starting an agent project in Python or TypeScript with LangChain tools
Anyone unsure whether a task needs LangChain, LangGraph or Deep Agents
Teams who want their coding agent to read current docs instead of relying on training data

Install ecosystem-primer

Claude Code

Copy this prompt.

Install the ecosystem-primer 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.

Pick one route. The README routes install all skills in the repo (the plugin, or the skills CLI with --skill '*'). SKILL.md says the layer-specific skills are required after the primer, so installing only ecosystem-primer is not enough. Set the LangSmith variables and a provider key before you start. The README does not document a Claude app upload for this repo, so there is no Claude app tab.

From the project README. Source.

What's in the skill

When it loads

INVOKE FIRST for any LangChain / LangGraph / Deep Agents agent building project before consulting other skills or writing any agent code. Required starting point for up to date info on framework selection (LangChain vs LangGraph vs Deep Agents vs hybrid composition), agent patterns, install, environment setup, and which skill to load next.

Sections in SKILL.md

Step 1 — Choose Your Tool
Tool Profiles
LangChain — agent framework
LangGraph — agent runtime
Deep Agents — agent harness
Mixing Layers
Step 2 — Set Environment Variables
Step 3 — How the Docs Work
Canonical landing pages
Accessing docs in an agent context
Step 4 — Load the Right Skill Next
LangChain
LangGraph
Deep Agents

Key facts

Author: LangChain Inc. (the vendor's own repo; plugin.json names LangChain as author)
License: MIT (LICENSE file and plugin.json agree)
Installs on skills.sh: 8,393 (Nov 2, 2026)
GitHub stars: 1,320 for the whole langchain-ai/langchain-skills repo (Nov 2, 2026)
Last repo commit: Oct 29, 2026
What it includes: 1 file: SKILL.md; no scripts
Needs: LANGSMITH_API_KEY, LANGSMITH_TRACING and LANGSMITH_PROJECT, plus a model provider key such as ANTHROPIC_API_KEY or OPENAI_API_KEY
Part of: langchain-skills plugin: the README says 21 skills, the repo tree has 22 skill folders

What it includes

A four-rule decision table: Deep Agents, LangGraph, LangChain agent, or plain model call
Rules are checked in order. Planning, files, memory or subagents point to Deep Agents. Custom control flow points to LangGraph. A fixed-tool agent points to LangChain create_agent.
Try: “I want an agent that researches a topic and writes a report; which LangChain tool should I use?”
Step 1: Choose your tool
Best-for and not-ideal-when lists for LangChain, LangGraph and Deep Agents
Each profile names its entry call: create_agent, StateGraph or create_deep_agent.
Tool profiles
Patterns for putting a LangGraph graph inside a Deep Agents orchestrator
A compiled graph can be registered as a named subagent that the orchestrator calls through the task tool.
Mixing layers
Sets LANGSMITH_API_KEY, LANGSMITH_TRACING and LANGSMITH_PROJECT for tracing
The skill says older LangSmith variable names no longer work. Provider keys are set as your stack needs.
Network · Step 2: Environment variables
Reads docs.langchain.com through the docs MCP server or the llms.txt index
Prefers live docs over training data. Landing pages are listed for LangChain, LangGraph, Deep Agents and LangSmith.
Network · Step 3: How the docs work
Names the quickstart or layer skill to load before any agent code
For example langchain-fundamentals, langgraph-fundamentals, or deep-agents-core first for Deep Agents.
Step 4: Load the next skill

From the official docs

Permissions

Low risk

What it can do

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

Watch out for

It tells you to turn LangSmith tracing on
The skill says to always set LANGSMITH_TRACING=true for observability. Decide whether you want tracing on before you use it with private data. github.com
The plugin route installs the whole set
plugin.json points at config/skills, which holds 22 skill folders in the git tree. Some, like swarm and eval-engineering, bundle scripts. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Talks to	docs.langchain.com (llms.txt and doc pages). LangSmith (the tracing variables point your agent at it)
Sign-in and keys	SKILL.md names LANGSMITH_API_KEY as an environment variable. It does not say where the key is stored.
Scopes	Tells the agent to set LANGSMITH_API_KEY, LANGSMITH_TRACING and LANGSMITH_PROJECT Tells the agent to fetch docs.langchain.com pages The plugin route installs every skill in the repo, including skills that bundle TypeScript and Python scripts

Questions

What does ecosystem-primer do?

It makes your coding agent pick the right LangChain tool for the job before writing code. It compares LangChain, LangGraph and Deep Agents, sets LangSmith tracing variables, and says which skill to load next.

Is it official?

Yes. It lives in the langchain-ai/langchain-skills repo, which is LangChain's own, and plugin.json names LangChain as author.

How do I install it?

Use the install box above. The README documents a Claude Code plugin and a skills CLI command that installs all the skills in the repo. SKILL.md says the layer-specific skills are required after the primer, so the whole set is the working install.

Does it need an API key?

The skill tells your agent to set LANGSMITH_API_KEY, LANGSMITH_TRACING and LANGSMITH_PROJECT for tracing, and a model provider key such as ANTHROPIC_API_KEY or OPENAI_API_KEY as your stack needs. SKILL.md says to copy the LangSmith names as written.

Why does it keep telling the agent to load another skill?

SKILL.md says the primer alone does not carry the current API. Step 4 lists the skill to load for each layer, such as langchain-fundamentals, langgraph-fundamentals or deep-agents-core.

Is it only for Claude Code?

No. The README says the skills install through npx skills for any agent that supports the Agent Skills specification and shows Claude Code, Codex and GitHub Copilot commands. The Claude app is not named, so the install box has no Claude app tab.

How is it different from the LangChain docs?

The docs hold the full reference. This skill is a short guide for choosing a layer and finding the right docs page, and it tells the agent to fetch live docs rather than rely on training data.

Is it safe?

It bundles no scripts. It tells your agent to set tracing variables, including an API key, and to fetch docs from docs.langchain.com. Our advice: keep the key in your environment or secrets store, not in code you commit.

Information

updated Nov 2, 2026

Repository: langchain-ai/langchain-skills
Folder: config/skills/ecosystem-primer

langchain-ai/langchain-skills skills.sh

Sources

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

Show all 4

README.md · github.com
marketplace.json · github.com
plugin.json · github.com
SKILL.md · github.com