Jupyter MCP Server
Jupyter Notebook management and real-time execution control for AI agents.
Data last scanned today · Reviewed today
Overview
Jupyter MCP Server provides a bridge between AI models and Jupyter Notebook environments. It allows agents to perform real-time notebook management, handle cell execution, and process output feedback to iterate on code failures dynamically.
Our verdict
With a security score of 53/100, this server is classified as high risk. The rating reflects that it is a community-maintained project rather than an official Jupyter release and lacks an editorially reviewed security audit for its authentication methods and read-only modes. It is suitable for experimental environments and local development but should be used with caution in sensitive or shared compute contexts.
- Zero direct dependencies reduce the supply-chain attack surface.
- Active development with commits within the last 24 hours.
- Supported by 34 contributors and over 1,200 GitHub stars.
- Released under the permissive BSD-3-Clause license.
- Not maintained by the official Jupyter vendor organization.
- Lacks external review for authentication and read-only mechanisms.
- Does not currently support external sandbox providers.
Tools
| Tool | Description | Risk |
|---|---|---|
| list_files | List files and directories in the Jupyter server's file system. | low |
| list_kernels | List all available and running kernel sessions on the Jupyter server. | low |
| launch_sandbox | Launch a code sandbox (eval/docker/jupyter/datalayer/kaggle/colab/monty/modal) as an alternative execution backend for execute_code. | medium |
| list_sandboxes | List launched code sandboxes and their state (active flag, variant, status, and selected code sandbox options). | low |
| use_sandbox | Select or clear the active sandbox used by execute_code, enabling dynamic routing between kernel-backed and sandbox-backed execution. | medium |
| terminate_sandbox | Stop and unregister a launched code sandbox. | high |
| connect_to_jupyter | Connect to a Jupyter server dynamically without restarting the MCP server. | medium |
| use_notebook | Connect to a notebook file, create a new one, or switch between notebooks. | medium |
| list_notebooks | List all notebooks available on the Jupyter server and their status | low |
| restart_notebook | Restart the kernel for a specific managed notebook. | medium |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What deployment types are supported for this MCP server?
This server is compatible with any Jupyter deployment, including local installations and JupyterHub, as well as Datalayer hosted Notebooks.
›What are the requirements for using sandbox-related tools like launch_sandbox or use_sandbox?
Using sandbox management tools requires the installation of the jupyter_mcp_sandboxes extension.
›What terminology change should I be aware of in version 1.2.0?
The component previously referred to as a Runtime or Jupyter Kernel is now named Code Sandbox.
›Are external sandbox providers supported?
No, this server does not support external sandboxes such as Kaggle, Google Colab, Modal, or others.
Alternatives
Changelog
- NewRelease 1.2.0
2026-07-31
- ImprovedRelease 1.1.4
2026-07-29
- FixRelease 1.1.3
2026-07-28
- FixRelease 1.1.2
2026-07-26
- FixRelease 1.1.1
2026-07-25
- FixRelease 1.1.0
2026-07-24
- FixRelease 1.0.7
2026-07-23
- FixRelease 1.0.6
2026-07-19
- FixRelease 1.0.5
2026-07-17
- FixRelease 1.0.4
2026-07-13
Badge
Maintain this server? Add the live badge to your README.