Jupyter FAQs

Question 1

What is Jupyter MCP Server?

Accepted Answer

Jupyter MCP Server is a tool that allows Model Context Protocol (MCP) clients, like Claude Desktop, to interact with Jupyter notebooks running in JupyterLab. It provides functionalities for cell management, execution, and file system access.

Question 2

What are the key features of Jupyter MCP Server?

Accepted Answer

Key features include connecting MCP clients to JupyterLab, managing notebook cells (add, delete, move, edit), executing cells and retrieving output, file system interaction, and kernel introspection/package management.

Question 3

What are the requirements for using Jupyter MCP Server?

Accepted Answer

You'll need Python (>= 3.10), JupyterLab, the `jupyter_collaboration` extension (version 2.0.1), `datalayer_pycrdt`, Docker, Pillow, and an MCP client like Claude Desktop. A dedicated Conda environment is strongly recommended.

Question 4

How do I configure Jupyter MCP Server to work with Claude Desktop?

Accepted Answer

You need to modify the `claude_desktop_config.json` file to include the `mcpServers` block with the correct Docker command and environment variables, including your JupyterLab server URL and token.

Question 5

What kind of file system operations are available?

Accepted Answer

You can list directory contents and get file content. For images, the server can resize them before returning them as base64 Data URIs. Binary files are simply described.

Jupyter

概要

主な機能

ユースケース