Robust new jupyter server
About
I was facing issues with https://github.com/datalayer/jupyter-mcp-server
Details
- Author
- kshitijdesai99
- Downloads
- 302
- Categories
- Search
Jump to
- Docker-based client for easy deployment
- Supports multi-collaborative Jupyter notebooks
- Integrates with Claude Desktop via MCP
- Token-based authentication with Jupyter server
- Uses jupyter-collaboration and pycrdt for real-time sync
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
Robust new jupyter serverCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
Set up a Python 3.12.10 environment, install dependencies (jupyterlab, jupyter-collaboration, etc.), start a Jupyter server with a token, build the Docker image named notebook-client, and add the appropriate configuration to your Claude Desktop config.json. Key commands include jupyter lab --port 8888 --IdentityProvider.token MY_TOKEN --ip 0.0.0.0 and docker build -t notebook-client ..
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"robust new jupyter server": {
"jupyter_mcp_server_kshitij": {
"command": "uv",
"args": [
"pip",
"install",
"jupyterlab",
"jupyter-collaboration",
"ipykernel",
"pycrdt"
]
}
}
}
}McpServers
{
"jupyter_mcp_server_kshitij": {
"command": "uv",
"args": [
"pip",
"install",
"jupyterlab",
"jupyter-collaboration",
"ipykernel",
"pycrdt"
]
}
}Jupyter MCP Server Client
This repository contains a Docker-based client for connecting to a Jupyter server with multi-collaborative protocol support.
Prerequisites
- Python 3.12.10
- Docker
- Git
Setup Instructions
1. Set Up Python Environment
# Create and activate virtual environment
python3.12 -m venv env
source env/bin/activate # Windows: env\Scripts\activate
Install package manager
pip install uv
Install required packages
uv pip install jupyterlab jupyter-collaboration ipykernel pycrdt
uv pip install jupyter-nbmodel-client jupyter_kernel_client
Fix dependencies (required for compatibility)
pip uninstall -y jupyter_server_ydoc pycrdt
uv pip install jupyter_server_ydoc pycrdt
2. Create Your Notebook
# Clone the repository -> Required
git clone https://github.com/kshitijdesai99/jupyter_mcp_server_kshitij.git
cd jupyter_mcp_server_kshitij
touch notebook.ipynb
3. Start Jupyter Server
jupyter lab --port 8888 --IdentityProvider.token MY_TOKEN --ip 0.0.0.0
> Note: Replace MY_TOKEN with your preferred authentication token or just keep it as it is.
4. Create a new notebook in case you get tornado.websocket.WebSocketClosedError (for Mac)
just delete the notebook.ipynb and then inside the jupyter server click on create a new notebook and rename it to notebook.ipynbDocker Instructions
Build the Docker Image
docker build -t notebook-client .
Run the Client Container --> just for testing not required for mcp
docker run \
-e NOTEBOOK_PATH="notebook.ipynb" \
-e SERVER_URL="http://host.docker.internal:8888" \
-e TOKEN="MY_TOKEN" \
notebook-client
> Note: When running from a different directory, use the full path to your notebook file.
Claude Desktop Integration
Add this configuration to your Claude desktop config.json file:
"jupyter": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"-e", "NOTEBOOK_PATH=notebook.ipynb",
"-e", "SERVER_URL=http://host.docker.internal:8888",
"-e", "TOKEN=MY_TOKEN",
"notebook-client"
]
}
> Important: no need to update notebook path
Troubleshooting
- Ensure the Jupyter server is running before connecting with the client
- Verify that your token matches between the server and client configuration
- For Windows paths in Docker, use double backslashes as shown in the example
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
