search_kaggle_datasetsLABSCO SUMMARY
From is there data on this to a downloaded folder and a first analysis, without leaving the chat.
Search returns the fields that decide whether a dataset is worth using, download count, last updated and usability rating, so the choice happens before anything is downloaded. The EDA piece is a prompt rather than an analysis: it hands your model a structured starting point and the model writes the code.
What it is
A server over the Kaggle API that searches datasets, downloads and unzips them locally, and generates an exploratory-analysis prompt for one.
What you get
- Dataset search by keyword, returning up to 10 results as JSON
- Each result carrying ref, title, subtitle, download count, last updated and usability rating
- Download and unzip of a dataset by its owner/dataset-slug reference, for example kaggle/titanic
- A destination path of your choosing, defaulting to ./datasets/<dataset_slug>/
- A generated EDA prompt covering data loading, missing-value checks, visualisations and summary statistics
Requirements
Python 3.10 or newer, a Kaggle account and an API token. Credentials come either from KAGGLE_USERNAME and KAGGLE_KEY or from a kaggle.json in the standard location. Runs locally, in Docker, or through Smithery.
Setup effort
Build from source — clone the repository, then follow the project's own run instructions
USE CASES
01
Finding out whether public data exists for a question
Search by keyword and read the ten results back with their titles and subtitles.
02
Comparing candidate datasets before committing to one
Weigh download count, last updated and usability rating against each other rather than opening ten Kaggle tabs.
03
Getting a dataset onto your disk to work with
Download and unzip it by reference, into the folder you name.
04
Starting the analysis of a dataset you just downloaded
Generate the EDA prompt and let your model write the loading, missing-value and plotting code from it.
2 tools
Listed in project docs. Your client may show a different list.
download_kaggle_datasetBEFORE YOU INSTALL
What to know first
Tools you get
2 tools named in the project's documentation.
Requires
- uv on PATH
- Python 3.10+
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
uv run kaggle-mcpMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"kaggle-mcp": {
"command": "uv",
"args": [
"run",
"kaggle-mcp"
],
"cwd": "<YOUR_PATH_TO_KAGGLE_MCP>",
"env": {
"KAGGLE_USERNAME": "<YOUR_KAGGLE_USERNAME>",
"KAGGLE_KEY": "<YOUR_KAGGLE_KEY>"
}
}
}
}
