Jupyter

Jupyter MCP Server provides an MCP interface to interact with Jupyter notebooks running in a local JupyterLab instance. It lets you add and execute code cells or insert Markdown content through MCP clients, enabling seamless notebook automation from other tools and workflows.

How to use

You run JupyterLab locally and expose it to the MCP server, then connect an MCP client to manage notebooks remotely. Use the client to send commands that add and execute code cells or insert Markdown cells into a notebook. When you start JupyterLab, you provide a token for authentication and expose the server on a network accessible address. The MCP server runs in Docker and communicates with your local JupyterLab instance to perform requested actions.

Typical usage pattern:

• Start JupyterLab with a secure token and accessible IP so the MCP server can reach it.
• Run the MCP server via Docker, supplying the Jupyter URL, token, and the path to your notebook.
• Use a client to call the available tools and modify notebooks programmatically.

How to install

Prerequisites you need before installation:

• Docker installed on your machine
• Access to the JupyterLab instance you will connect to (host URL and a valid token)
• A notebook path you want to modify (relative to where JupyterLab was started)

Install and run the MCP-enabled Jupyter server with Docker using the following configuration snippets. There are two environment setups depending on your operating system. Use the one that matches your environment.

{
  "mcpServers": {
    "jupyter": {
      "type": "stdio",
      "name": "jupyter_mcp",
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "SERVER_URL",
        "-e",
        "TOKEN",
        "-e",
        "NOTEBOOK_PATH",
        "--network=host",
        "datalayer/jupyter-mcp-server:latest"
      ],
      "env": {
        "SERVER_URL": "http://localhost:8888",
        "TOKEN": "MY_TOKEN",
        "NOTEBOOK_PATH": "notebook.ipynb"
      }
    }
  }
}

Notes

Security: keep the token secret and limit exposure of the JupyterLab URL. The MCP server assumes the JupyterLab interface is reachable at the SERVER_URL you provide. If you run in a network-isolated environment, ensure the Docker container can reach the host JupyterLab instance.

Additional configuration and troubleshooting

If you encounter connectivity issues, verify that the JupyterLab server is accessible at the URL you provided and that the token in TOKEN matches what JupyterLab expects. Ensure the notebook path exists relative to where JupyterLab was started. When running on Linux, you may prefer --network=host to simplify networking between Docker and the host.

Available tools

add_execute_code_cell

Add and execute a code cell in a Jupyter notebook using MCP client input and trigger execution in the notebook.

add_markdown_cell

Add a Markdown cell to a Jupyter notebook using MCP client input.