mcpField
L
langgraph-python-quickstart

langgraph-python-quickstart

github.com/langchain-aiAgent workflows1.3k5.3k installsUpdated 3d ago
Repository

About langgraph-python-quickstart

A quickstart skill for new LangGraph Python projects. It directs implementation to the current official docs and a calculator agent example, with minimal local setup and a model chosen by the user.

Key features

  1. 01Ask for a model provider and model
  2. 02Prefer the Graph API
  3. 03Create the project in a new directory
  4. 04Configure only the model API key
  5. 05Run the example and show its output

Use cases

  1. 01Start a LangGraph Python agent project
  2. 02Configure a chat model chosen by the user
  3. 03Check that a calculator agent handles a request

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/langchain-ai/langchain-skills --skill langgraph-python-quickstart

After installing, type/langgraph-python-quickstartin Claude Code to confirm it loaded.

Details

From the project README

LangGraph Python quickstart

Follow the live docs — do not invent an alternate API from memory:

https://docs.langchain.com/oss/python/langgraph/quickstart

Fetch that page (Docs MCP or HTTP) and implement what it shows (calculator / math agent with the Graph API). Prefer the Graph API path over the Functional API unless the user asks otherwise. Skip IPython graph visualization.

Local setup constraints

Apply these on top of the quickstart (they keep setup minimal and model-agnostic):

  1. Ask which provider/model to use. Showcase that LangGraph works with any LangChain chat model. Suggested prompt:

    Which model should this agent use? Pass a provider:model string — e.g. openai:gpt-5.5, anthropic:claude-sonnet-5, google_genai:gemini-2.5-flash-lite. Default if you're unsure: anthropic:claude-sonnet-5.

    The docs often hardcode Anthropic — replace with init_chat_model("<MODEL>") (or equivalent) using their choice. If using Claude Sonnet 5+, omit temperature / top_p / top_k (unsupported).

  2. Create a new directory (e.g. langgraph-agent/) and do all work there — do not pollute the open project.

  3. Only secret: the provider API key in .env (gitignored). No LangSmith / Tavily unless they ask. Prefer they edit .env themselves — don't paste keys into chat.

  4. Install packages from the quickstart plus the provider package for their model.

  5. Run the example (e.g. “Add 3 and 4.”), show output, then stop. Point to langgraph-fundamentals for next steps. For a higher-level agent API, use LangChain create_agent instead.

Comments

Markdown supported