langgraph-fundamentals Skill for Claude Code: Basics

Teaches your agent the core of LangGraph: StateGraph, state and reducers, nodes, edges, Command, Send, streaming and error handling.

Installs: 18.5K; on skills.sh
Stars: 1.3K; on GitHub
Files: 2; besides SKILL.md
Updated: Oct 2026; last repo commit

The skill's description says to invoke it when writing any LangGraph code. It gives the agent a five-step method for designing a graph and short tables for picking an edge type, a state update strategy, a stream mode and an error handling approach. Python and TypeScript details sit in two reference files that the agent reads only when it writes code.

It flags two traps. A Command with goto adds a dynamic edge, and static edges still run, so both targets execute. A list field without a reducer is overwritten and not appended. It bundles no scripts. It is part of LangChain's skills repo, which the README marks as early development.

Good for

Python or TypeScript developers building agent workflows with LangGraph
Developers debugging graphs that overwrite state or run extra nodes
Teams choosing between LangGraph, LangChain agents and Deep Agents

Install langgraph-fundamentals

Claude Code

Copy this prompt.

Install the langgraph-fundamentals skill in Claude Code. Run this command:

npx skills add https://github.com/langchain-ai/langchain-skills --skill langgraph-fundamentals

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.

From its listing on skills.sh. Source.

What's in the skill

When it loads

INVOKE THIS SKILL when writing ANY LangGraph code. Covers StateGraph, state schemas, nodes, edges, Command, Send, invoke, streaming, and error handling.

Sections in SKILL.md

Designing a LangGraph application
State Management
Nodes
Edges
Command
Send API
Running Graphs: Invoke and Stream
Error Handling
Core boundaries
Implementation references

Key facts

Author: langchain-ai
Repo: langchain-ai/langchain-skills, folder config/skills/langgraph-fundamentals
Installs on skills.sh: 18,457 as of 2026-11-02
GitHub stars: 1,320 for the whole langchain-ai/langchain-skills repo
License: MIT (as detected by GitHub)
Last repo commit: 2026-10-29
What is inside: 3 files including SKILL.md; 2 reference files (Python and TypeScript), no scripts
skills.sh audits: Gen Agent Trust Hub: pass; Socket: pass; Snyk: pass

What it includes

Five steps from flowchart to compiled graph
Map the steps, decide what each step does, design shared state, build the nodes and wire them with edges. Graphs must be compiled before they run.
Read · Design a LangGraph application
Overwrite, append or custom reducer
No reducer means a field is overwritten. A list that should grow needs an append reducer such as operator.add or a custom function.
Write · State management
Static, conditional, Command or Send
Use add_edge for a fixed flow, conditional edges for routing on state, Command to update state and route together, and Send to fan out to parallel workers.
Write · Edge selection
Update, goto and resume in one return
Command combines a state update, the next node and a resume value after an interrupt. Its return type declares which nodes it can go to.
Write · Command
Fan out to parallel workers
Return Send items from a conditional edge to start workers. The results field needs a reducer or the last worker wins.
Write · Send API
Four stream modes
invoke runs to completion. Stream modes are values (full state), updates (deltas), messages (LLM tokens) and custom (your own data).
Write · Invoke and stream
Match the error type to the handler
Transient errors use a retry policy, tool failures go back to the LLM, missing user information uses interrupt, and unexpected errors are left to surface.
Write · Error handling
Loaded only when code is written
references/python.md and references/typescript.md hold the code for each topic. The agent reads one, or both only if the task covers both languages.
Read · Python and TypeScript references

From the official docs

Questions

What does langgraph-fundamentals do?

It teaches your agent how to write LangGraph code. The core ideas are StateGraph, state with reducers, nodes, edges, Command, Send, streaming and error handling, in both Python and TypeScript.

How do I install it?

Use the install box at the top of this page. The LangChain README also gives an npx skills command that installs every skill in the repo, which is more than this skill needs.

Does it need an API key?

No. The skill is guidance only and names no API key or service. The code it helps you write may call model providers that have their own keys.

Does it cover saving state between runs?

No. Checkpointers, thread IDs, time travel and the Store are in the separate langgraph-persistence skill from the same repo. Human-in-the-loop with interrupt() is covered by a human-in-the-loop skill that this one points to.

Which languages does it support?

Python and TypeScript. It has one reference file for each and reads the one that matches your project.

Is it safe?

It bundles no scripts and tells the agent how to write LangGraph code. Review the code it writes like any other generated code.

Information

updated Nov 2, 2026

Repository: langchain-ai/langchain-skills
Folder: config/skills/langgraph-fundamentals

langchain-ai/langchain-skills skills.sh

Sources

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

Show all 4

langgraph-fundamentals SKILL.md · github.com
langchain-ai/langchain-skills README · github.com
langgraph-fundamentals on skills.sh · skills.sh
langchain-ai/langchain-skills · github.com