Kaggle MCP Server

A full-featured Model Context Protocol (MCP) server for the Kaggle API — 96 tools across competitions, datasets, kernels, models, benchmarks, discussions, and workflow utilities.

Installation

pip install kaggle-mcp-server

Prerequisites

1. Kaggle API credentials — place your kaggle.json at ~/.kaggle/kaggle.json:

{"username":"YOUR_USERNAME","key":"YOUR_API_KEY"}

Get your API key from kaggle.com/settings → "Create New Token".

2. Python 3.12+

Usage

With Cursor IDE

Add to .cursor/mcp.json:

{
  "mcpServers": {
    "kaggle": {
      "command": "kaggle-mcp-server"
    }
  }
}

With Claude Desktop

Add to your Claude Desktop config:

{
  "mcpServers": {
    "kaggle": {
      "command": "kaggle-mcp-server"
    }
  }
}

Standalone

kaggle-mcp-server

Tools (96 total)

Competitions (16 tools)

Tool	Description
competitions_list	Search and list competitions
competition_get	Get detailed competition info
competition_files	List competition data files
competition_tree_files	Hierarchical tree view of competition data
competition_download	Download competition data (returns URL)
competition_download_single_file	Download a single competition file locally
competition_submit	Submit predictions via blob token
submit_local_file	Submit a local prediction file
submit_code_competition	Submit to code competitions
competition_submissions	View submission history
competition_get_submission	Get single submission details
submission_score	Get/poll submission score
competition_leaderboard	View top 20 leaderboard
competition_leaderboard_download	Download full leaderboard
leaderboard_position	Find a team/user's rank
competition_data_summary	Get data files summary

Competition Workflow (5 tools)

Tool	Description
setup_comp	Download and extract competition data locally
competition_full_setup	One-shot setup: info + download + preview
upcoming_deadlines	Show competitions with closest deadlines
my_competitions	List competitions you've entered
competition_top_kernels	Top public notebooks for a competition

Datasets (20 tools)

Tool	Description
datasets_list	Search and list datasets
dataset_get	Get full dataset info
dataset_files	List files in a dataset
dataset_tree_files	Hierarchical tree view of dataset files
dataset_files_summary	Get file count and total size
dataset_download	Download dataset (returns URL)
dataset_download_file	Download a single dataset file
download_dataset_local	Download and extract dataset locally
dataset_metadata	Get dataset metadata
dataset_update_metadata	Update title/description/license
dataset_create	Create dataset via blob tokens
create_dataset_from_files	Create dataset from local directory
dataset_create_version	Create new version via tokens
push_dataset_version	Push new version from local directory
dataset_delete	Delete a dataset
dataset_status	Check dataset processing status
file_upload	Upload file and get blob token
my_datasets	List your datasets
datasets_by_user	List datasets by a specific user
check_dataset_exists	Check if a dataset exists

Kernels / Notebooks (15 tools)

Tool	Description
kernels_list	Search and list notebooks
kernel_pull	Get notebook source code
kernel_push	Push/save a notebook
push_notebook_file	Push local .ipynb to Kaggle
kernel_output	Download kernel output (URL)
kernel_download_output_zip	Download kernel output locally
kernel_status	Check kernel execution status
kernel_files	List kernel files
kernel_delete	Delete a kernel
kernel_initialize	Initialize kernel template locally
kernel_session_create	Create interactive session
kernel_session_status	Get session status
kernel_session_output	List session output files
kernel_session_cancel	Cancel running session
generate_notebook_metadata	Generate kernel-metadata.json

Models (16 tools)

Tool	Description
models_list	Search and list models
model_get	Get model details
model_create	Create a new model
model_update	Update model info
model_delete	Delete a model
model_metrics	Get model performance metrics
model_instances_list	List model instances
model_instance_get	Get instance details
model_instance_create	Create a new instance
model_instance_delete	Delete an instance
model_instance_files	List instance files
model_instance_versions	List instance versions
model_instance_version_create	Create new version
model_instance_version_download	Download version files
model_instance_version_files	List version files
model_instance_version_delete	Delete a version

Discussions (10 tools)

Tool	Description
discussions_search	Search discussions
discussions_list	List discussions for competition/dataset
discussion_detail	Get discussion content
discussion_comments	Get discussion comments
discussion_comments_search	Search across all comments
discussions_by_source	Browse by source type
discussions_solutions	Browse competition solutions
discussions_writeups	Browse write-ups by type
discussions_trending	Browse trending discussions
discussions_my	List your discussions

Benchmarks (1 tool)

Tool	Description
benchmark_leaderboard	Get benchmark leaderboard

Data Preview & Analysis (4 tools)

Tool	Description
preview_csv	Preview first N rows of a CSV
preview_data_file	Preview any text data file
csv_column_analysis	Analyze column types and stats
compare_csvs	Diff two CSV files

Workflow Utilities (9 tools)

Tool	Description
generate_starter_notebook	Auto-generate competition starter notebook
search_everything	Unified search across competitions, datasets, notebooks
list_local_files	List local files with sizes
track_operation	Monitor long-running Kaggle operations
download_pip_library	Download pip wheels for offline use
download_pip_requirements	Download requirements.txt wheels
create_library_dataset	Upload pip library as Kaggle dataset
create_requirements_dataset	Upload requirements as Kaggle dataset
get_local_library_version	Check local wheel version

License

MIT