kaggle-mcpMCP server for Kaggle

— (0)

Reviewed by Daniel Nikulshyn·Updated July 2026

Overview

Kaggle MCP Server is a Model Context Protocol (MCP) server that exposes Kaggle dataset search, download, and EDA prompt generation to MCP clients. It allows users to search Kaggle datasets by keyword, download and unzip datasets locally, and generate starter Exploratory Data Analysis (EDA) prompts for Kaggle datasets. The server supports Kaggle credentials via environment variables or the standard kaggle.json file and can run locally, in Docker, or through Smithery.

Key features

Search Kaggle datasets by keyword
Download and unzip Kaggle datasets locally
Generate starter EDA prompts for Kaggle datasets
Support for Kaggle credentials via environment variables or kaggle.json

Pricing

Model: Free
Category: MCP Servers
Rating: No reviews yet

Use cases

Dataset Search

Use the Kaggle MCP Server to search for datasets related to a specific topic, such as 'heart disease'.

Dataset Download and EDA

Download a Kaggle dataset and generate an EDA prompt to quickly get started with data analysis.

Pros & Cons

Pros

Exposes Kaggle dataset search, download, and EDA prompt generation to MCP clients
Supports Kaggle credentials via environment variables or kaggle.json file
Can run locally, in Docker, or through Smithery
Allows for easy integration with MCP-compatible clients like Claude Desktop

Cons

Requires Python 3.10+ and a Kaggle account with API token
Needs MCP-compatible client for functionality
Configuration can be complex for Docker and Smithery setups

Reviews

No reviews yet. Be the first!

Q&A

What functionality does the MCP server expose to compatible clients?

It offers three MCP capabilities: search_kaggle_datasets (to find up to 10 datasets by keyword), download_kaggle_dataset (to download and unzip a dataset locally), and generate_eda_notebook (to create a starter EDA prompt for a given dataset reference.

Asked by Sofia Lindqvist · Sep 28, 2025

Can the server be run in Docker or via Smithery, and what extra setup is needed?

Yes, the repository includes a Dockerfile for containerized deployment, and a smithery.yaml for Smithery. You’ll need to ensure your Kaggle credentials are accessible inside the container or Smithery environment, typically by mounting the .kaggle directory or providing env vars.

Asked by Piotr Baranowski · Aug 26, 2025

How do I provide my Kaggle credentials to the server?

You can set the KAGGLE_USERNAME and KAGGLE_KEY environment variables in a .env file, or place the kaggle.json file in the standard location (~/.kaggle/kaggle.json on macOS/Linux or C:\Users\<User>\.kaggle\kaggle.json on Windows).

Asked by Mia Andersen · Aug 22, 2025

What Python version and dependencies are required to run Kaggle MCP Server?

You need Python 3.10 or newer. Install the required packages via uv sync, uv run, or pip install -r requirements.txt as described in the repository.

Asked by Naomi Suzuki · Aug 11, 2025