kaggle-mcp-server

This project provides an MCP (Model Context Protocol) server for interacting with Kaggle from Claude Desktop (or any MCP-compatible client)

pip install -r requirements.txt

Go to your Kaggle account -> Settings

Scroll to API section → click Create New API Token. This will download a file called kaggle.json.

Place kaggle.json in the following location on your system:

Linux/Mac:

~/.kaggle/kaggle.json

Windows:

C:\Users\<your-username>\.kaggle\kaggle.json

To connect Claude with this MCP server, create or update the file:

Claude/claude_desktop_config.json

with the following contents:

{ "mcpServers": { "Kaggle": { "command": "<path-to-your-python-executable>", "args": ["<path-to-your-kaggle-server.py>"] } } }

Replace <path-to-your-python-executable> with the path to your Python interpreter (e.g. python or C:/Python/python.exe).

Replace <path-to-your-kaggle-server.py> with the path to this repo’s kaggle-server.py file.

Once configured:

Start Claude Desktop.

Start the Kaggle MCP server (Run kaggle-server.py as a Python File).

You can now use MCP tools defined in kaggle-server.py, for example:

get_competitions_list: Fetches the list of available Kaggle competitions and returns them in a JSON-friendly format.

kaggle-server.py registers MCP tools using the @mcp.tool() decorator.

The server communicates with Claude via MCP and exposes Kaggle API methods.

Example tool:

@mcp.tool() def get_competitions_list() -> list[dict]: """ Fetch the list of available Kaggle competitions.