Jupyter MCP Server

What it does

Connects AI assistants to Jupyter Notebooks in real-time, allowing the AI to read cells, execute code, and manage notebook state. It supports multimodal outputs, meaning the AI can see plots and images generated by the kernel.

Tools

• execute_cell: Runs a specific notebook cell and returns output.
• execute_code: Executes arbitrary code directly in the kernel.
• read_notebook: Reads source content of notebook cells.
• insert_cell: Adds new code or markdown cells to a notebook.
• list_notebooks: Lists all available notebooks on the server.
• use_notebook: Switches between different notebook files.
• connect_to_jupyter: Dynamically connects to a Jupyter server.

Installation

To use with Claude Desktop, add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "jupyter": {
      "command": "uvx",
      "args": ["jupyter-mcp-server@latest"],
      "env": {
        "JUPYTER_URL": "http://localhost:8888",
        "JUPYTER_TOKEN": "YOUR_TOKEN",
        "ALLOW_IMG_OUTPUT": "true"
      }
    }
  }
}

Supported hosts

• Claude Desktop
• Cursor
• Windsurf
• Gemini CLI