Jupyter MCP Server

Developmentstdio

Open-source MCP server for working with Jupyter notebooks and JupyterHub. Lets agents list and switch notebooks, read and edit cells, execute code, inspect kernel output, and return multimodal results such as plots and images. Supports local Jupyter deployments as well as optional sandbox backends for isolated code execution.

jupyter notebooks data-science code-execution python multimodal

Connect

Add this configuration to .claude/mcp.json

{
  "mcpServers": {
    "jupyter": {
      "command": "uvx",
      "args": [
        "jupyter-mcp-server@latest"
      ],
      "env": {
        "JUPYTER_URL": "<YOUR_JUPYTER_URL>",
        "JUPYTER_TOKEN": "<YOUR_JUPYTER_TOKEN>",
        "ALLOW_IMG_OUTPUT": "<YOUR_ALLOW_IMG_OUTPUT>"
      }
    }
  }
}

Tools (6)

list_notebooks

List notebooks available on the connected Jupyter server and report their status.

{
  "type": "object"
}

use_notebook

Connect to, create, or switch the active Jupyter notebook.

{
  "type": "object"
}

read_notebook

Read notebook cell source content in a brief or detailed representation.

{
  "type": "object"
}

insert_cell

Insert a code or Markdown cell at a specified notebook position.

{
  "type": "object"
}

execute_cell

Execute a notebook cell and return text, image, plot, or other supported output.

{
  "type": "object"
}

execute_code

Execute code in the active Jupyter kernel or selected sandbox backend.

{
  "type": "object"
}

Resources

This server does not expose any resources.

Prompts (1)

jupyter-cite

Cite specific cells from a selected notebook in an MCP client conversation.

Server information

Author: Datalayer (@datalayer)
Repository: https://github.com/datalayer/jupyter-mcp-server
License: BSD-3-Clause
Stars: 0
Last updated: August 22, 2026