jupyter

This MCP server provides real-time, MCP-compatible control over Jupyter notebooks. It exposes a rich set of tools that let clients list and manage files, kernels, and notebooks, as well as perform per-cell operations and notebook-level actions. Notably, you can use server management tools like list_files, list_kernels, and connect_to_jupyter to discover and switch between Jupyter sessions on the fly. For notebook management, tools such as use_notebook, list_notebooks, restart_notebook, and unuse_notebook help you attach to and switch between multiple notebooks, monitor their status, and manage their lifecycles. When performing code interactions, you can read_cell, insert_cell, delete_cell, overwrite_cell_source, and execute_cell, or run a code block directly with execute_code or insert_execute_code_cell. If you’re using JupyterLab, the server can integrate with jupyter-mcp-tools to expose additional UI-driven commands, such as notebook_run-all-cells and notebook_get-selected-cell.

Prerequisites:

• Python 3.8 or newer installed on your system
• pip (Python package installer)

1. Create a Python virtual environment (recommended): python3 -m venv venv source venv/bin/activate # On Windows use: venv\Scripts\activate

2. Install the MCP server package from PyPI: pip install jupyter-mcp-server

3. Run the MCP server locally (from the project root or anywhere accessible): python -m jupyter_mcp_server

4. Optional: configure environment variables or command-line options as needed per your deployment (see the project docs for details).

5. If you want to run via a container, pull the Docker image and run as documented in the repository's Docker usage section.

Tips and common considerations:

• The server is designed to be MCP-compatible and works with any MCP client (Claude Desktop, Cursor, Windsurf, etc.).
• When using JupyterLab mode, additional tools from jupyter_mcp_tools may be enabled by default; you can customize which tools are exposed via configuration (allowed_jupyter_mcp_tools).
• If you plan to deploy in a multi-notebook environment, take advantage of multi-notebook management tools (use_notebook, list_notebooks, restart_notebook, unuse_notebook).
• For dynamic Jupyter server changes, you can use connect_to_jupyter to switch contexts without restarting the MCP server.
• Ensure network access between the MCP server and your Jupyter deployments; firewall rules and proper port exposure are common sources of connectivity issues.