Jupyter Notebook

What is JupyterMCP?

JupyterMCP is a project that integrates Jupyter Notebook (version 6.x) with Claude AI via the Model Context Protocol (MCP). This allows Claude AI to directly interact with and control Jupyter Notebooks, facilitating AI-assisted code execution, data analysis, and visualization.

How to use JupyterMCP?

Prerequisites: Ensure you have Python 3.12+, uv package manager, and the Claude AI desktop application.
Installation: Clone the repository, create a virtual environment, install the jupyter-mcp kernel, and optionally install additional Python packages.
Claude Configuration: Configure Claude desktop integration by adding the jupyter MCP server to claude_desktop_config.json.
Starting Connection: Start your Jupyter Notebook (v6.x) server, create a new notebook with the jupyter-mcp kernel, and run the provided Python code to initialize the WebSocket server. Then, launch Claude desktop with MCP enabled.

Key Features of JupyterMCP

Two-way communication: Establishes a WebSocket-based server for communication between Jupyter Notebook and Claude AI.
Cell manipulation: Allows insertion, execution, and management of notebook cells.
Notebook management: Supports saving notebooks and retrieving notebook information.
Cell execution: Enables running specific cells or all cells in a notebook.
Output retrieval: Provides options to get output content from executed cells, including text and images.
Cell editing: Allows editing the content of existing cells.
Slideshow type setting: Enables setting the slideshow type for cells.

Use Cases of JupyterMCP

JupyterMCP can be used for various AI-assisted tasks within Jupyter Notebooks, such as:

Automated code generation and execution: Claude AI can generate and run Python or Stata code directly in the notebook.
Data analysis and visualization: AI can assist in performing data analysis and creating visualizations.
Interactive presentations: Create dynamic presentations with AI-generated content and appropriate slideshow types.
Educational purposes: Facilitate learning and experimentation with AI-driven coding and data science tasks.

FAQ about JupyterMCP

Compatibility: JupyterMCP is ONLY compatible with Jupyter Notebook version 6.x. It does not work with Jupyter Lab, Jupyter Notebook v7.x, VS Code Notebooks, or Google Colab.
Troubleshooting Connection Issues: If you encounter connection timeouts, the client has a reconnection mechanism. Restarting the WebSocket server can also help.
Cell Execution Problems: Ensure cell content is valid and the notebook kernel is running.
WebSocket Port Conflicts: The server automatically tries to find an available port if the default (8765) is in use.
Security Disclaimer: This is an experimental project that runs arbitrary Python code. Users should exercise caution and back up important data.

JupyterMCP - Jupyter Notebook Model Context Protocol Integration

JupyterMCP connects Jupyter Notebook to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Jupyter Notebooks. This integration enables AI-assisted code execution, data analysis, visualization, and more.

⚠️ Compatibility Warning

This tool is compatible ONLY with Jupyter Notebook version 6.x.

It does NOT work with:

Jupyter Lab
Jupyter Notebook v7.x
VS Code Notebooks
Google Colab
Any other notebook interfaces

Features

Two-way communication: Connect Claude AI to Jupyter Notebook through a WebSocket-based server
Cell manipulation: Insert, execute, and manage notebook cells
Notebook management: Save notebooks and retrieve notebook information
Cell execution: Run specific cells or execute all cells in a notebook
Output retrieval: Get output content from executed cells with text limitation options

Components

The system consists of three main components:

WebSocket Server (jupyter_ws_server.py): Sets up a WebSocket server inside Jupyter that bridges communication between notebook and external clients
Client JavaScript (client.js): Runs in the notebook to handle operations (inserting cells, executing code, etc.)
MCP Server (jupyter_mcp_server.py): Implements the Model Context Protocol and connects to the WebSocket server

Installation

Prerequisites

Python 3.12 or newer (probably also work with older versions, but not tested)
uv package manager
Claude AI desktop application

Installing uv

If you're on Mac:

brew install uv

On Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

For other platforms, see the uv installation guide.

Setup

Clone or download this repository to your computer:

git clone https://github.com/jjsantos01/jupyter-notebook-mcp.git

Create virtual environment with required packages an install jupyter-mcp kernel, so it can be recognized by your jupyter installation, if you had one before.
```
uv run python -m ipykernel install --name jupyter-mcp
```
(optional) Install additional Python packages for your analysis:
```
uv pip install seaborn
```

Configure Claude desktop integration: Go to Claude > Settings > Developer > Edit Config > claude_desktop_config.json to include the following:

   {
    "mcpServers": {
        "jupyter": {
            "command": "uv",
            "args": [
                "--directory",
                "/ABSOLUTE/PATH/TO/PARENT/REPO/FOLDER/src",
                "run",
                "jupyter_mcp_server.py"
            ]
        }
    }
}

Replace /ABSOLUTE/PATH/TO/ with the actual path to the src folder on your system. For example:

Windows: "C:\\Users\\MyUser\\GitHub\\jupyter-notebook-mcp\\src\\"
Mac: /Users/MyUser/GitHub/jupyter-notebook-mcp/src/

If you had previously opened Claude, then File > Exit and open it again.

Usage

Starting the Connection

Start your Jupyter Notebook (version 6.x) server:
```
uv run jupyter nbclassic
```
Create a new Jupyter Notebook and make sure that you choose the jupyter-mcp kernel: kernel -> change kernel -> jupyter-mcp
In a notebook cell, run the following code to initialize the WebSocket server:
```
import sys
sys.path.append('/path/to/jupyter-notebook-mcp/src')  # Add the path to where the scripts are located

from jupyter_ws_server import setup_jupyter_mcp_integration

# Start the WebSocket server inside Jupyter
server, port = setup_jupyter_mcp_integration()
```
Don't forget to replace here '/path/to/jupyter-notebook-mcp/src' with src folder on your system. For example:
- Windows: "C:\\Users\\MyUser\\GitHub\\jupyter-notebook-mcp\\src\\"
- Mac: /Users/MyUser/GitHub/jupyter-notebook-mcp/src/
Launch Claude desktop with MCP enabled.

Using with Claude

Once connected, Claude will have access to the following tools:

ping - Check server connectivity
insert_and_execute_cell - Insert a cell at the specified position and execute it
save_notebook - Save the current Jupyter notebook
get_cells_info - Get information about all cells in the notebook
get_notebook_info - Get information about the current notebook
run_cell - Run a specific cell by its index
run_all_cells - Run all cells in the notebook
get_cell_text_output - Get the output content of a specific cell
get_image_output - Get the images output of a specific cell
edit_cell_content - Edit the content of an existing cell
set_slideshow_type- Set the slide show type for cell

⚠️ DISCLAIMER

This is an experimental project and should be used with caution. This tool runs arbitrary Python code in your computer, which could potentially modify or delete data if not used carefully. Always back up your important projects and data.

Example Prompts

Ask Claude to perform notebook operations:

Python example

You can check the example notebook and the video demo

You have access to a Jupyter Notebook server.

I need to create a presentation about Python's Seaborn library.  
The content is as follows:

- What is Seaborn?
- Long vs. Wide data format
- Advantages of Seaborn over Matplotlib
- Commonly used Seaborn functions
- Live demonstration (comparison of Seaborn vs. Matplotlib)
  - Bar plot
  - Line plot
  - Scatter plot

For each concept, I want the main explanations provided in markdown cells, followed by one or more Python code cells demonstrating its usage. Keep the text concise—the cells shouldn't exceed 10 lines each.

Use appropriate slideshow types for each cell to make the presentation visually appealing.

Check Here the full conversation

Stata example

For this example, you need the Stata Software (v17 or later), which is not open source. If you already have Stata, you need to install the stata-setup package:

uv pip install stata-setup

Then, at the begining of your notebook, you need to additionally include:

import stata_setup
stata_setup.config('your_stata_installation_directory', 'your_stata_edition')

You can check the example notebook and the video demo

This exercise comes from Professor John Robert Warren webpage

You have access to a Jupyter Notebook server. By default it runs Python, but you can run Stata (v18) code in this server using the %%stata magic, for example:

%%stata
display "hello world"

Run the available tools to solve the exercise, execute the code, and interpret the results.

**EXERCISE:**

In this exercise, you will use data from the American Community Survey (ACS). The ACS is a product of the U.S. Census Bureau and involves interviewing millions of Americans each year. For an introduction to the ACS, visit the ACS website (here).

For this exercise, I have created a data file containing two variables collected from respondents of the 2010 ACS who lived in one of two metropolitan areas: Minneapolis/St Paul and Duluth/Superior. The two variables are: (1) People's poverty status and (2) the time it takes people to commute to work.

Use STATA syntax files you already have (from the first assignment or class examples) and modify them to accomplish the following goals.

1. Read the data file (`"./stata_assignment_2.dat"`) for this assignment into STATA.
2. Be sure to declare "zero" as a missing value for `TRANTIME`, the commuting time variable.
3. Create a new dichotomous poverty variable that equals "1" if a person's income-to-poverty-line ratio (`POVRATIO`) is less than 100, and "0" otherwise; see the bottom of the assignment for an example of how to do this in STATA.
4. Separately for Minneapolis/St Paul and Duluth/Superior, produce:
   - a histogram of the commuting time (`TRANTIME`) variable.
   - measures of central tendency and spread for commuting time.
   - a frequency distribution for the poverty status (0 vs 1) variable.
5. Separately for Minneapolis/St Paul and Duluth/Superior, use STATA code to produce:
   - a 95% confidence interval for the mean commuting time.
   - a 95% confidence interval for the proportion of people who are poor. See below for an example of how to do this in STATA.

Use the results from step #4 above to:

6. Separately for Minneapolis/St Paul and Duluth/Superior, manually calculate:
   - a 95% confidence interval for the mean commuting time.
   - a 95% confidence interval for the proportion of people who are poor.
7. Confirm that your answers from steps #5 and #6 match.

Based on the results above, answer this question:

8. How do you interpret the confidence intervals calculated in steps #5 and #6 above?

9. Finally, create a do file (.do) with the all the Stata code and the answers as comments.

---

**DESCRIPTION OF VARIABLES IN "STATA ASSIGNMENT 2.DAT"**

**METAREAD** (Column 4-7)  
Metropolitan Area  
- `2240`: Duluth-Superior, MN/WI  
- `5120`: Minneapolis-St. Paul, MN  

**POVRATIO** (Column 18-20)  
Ratio of person's income to the poverty threshold:  
- `<100`: Below Poverty Line  
- `100`: At Poverty Line  
- `>100`: Above Poverty Line  

**TRANTIME** (Column 21-23)  
Travel time to work  
- `0`: Zero minutes  
- `1`: 1 Minute  
- etc.

Check Here the full conversation

Testing with External Client

You can test the functionality without using Claude Desktop with the included external client:

uv run python src/jupyter_ws_external_client.py

This will provide an interactive menu to test some available functions.

For automated testing of all commands:

uv run python src/jupyter_ws_external_client.py --batch

Troubleshooting

Connection Issues: If you experience connection timeouts, the client includes a reconnection mechanism. You can also try restarting the WebSocket server.
Cell Execution Problems: If cell execution doesn't work, check that the cell content is valid Python/Markdown and that the notebook kernel is running.
WebSocket Port Conflicts: If the default port (8765) is already in use, the server will automatically try to find an available port.

Limitations

Only supports Jupyter Notebook 6.x
Text output from cells is limited to 1500 characters by default
Does not support advanced Jupyter widget interactions
Connection may timeout after periods of inactivity

License

MIT

Other Jupyter MCPs

This project is inspired by similar MCP integrations for Jupyter as: