MCP Servers

1. Go to https://www.kaggle.com/settings → API → Create New API Token
2. Set the environment variable:

export KAGGLE_API_TOKEN="KGAT_xxxxxxxxxxxx"

Download the token file from Kaggle settings, it will be saved to ~/.kaggle/kaggle.json:

{"username": "your_username", "key": "your_api_key"}

{
  "mcpServers": {
    "kaggle": {
      "command": "uvx",
      "args": ["kaggle-mcp-server"],
      "env": {
        "KAGGLE_API_TOKEN": "KGAT_xxxxxxxxxxxx"
      }
    }
  }
}

{
  "mcpServers": {
    "kaggle": {
      "command": "python",
      "args": ["-m", "kaggle_mcp.server"],
      "env": {
        "KAGGLE_API_TOKEN": "KGAT_xxxxxxxxxxxx"
      }
    }
  }
}

1. competitions_list — Search and list Kaggle competitions.

   • Inputs:
     • search (string, optional): Search term to filter competitions.
     • category (string, optional): Filter by category (e.g. featured, research, playground).
     • sort_by (string, optional): Sort order (latestDeadline, numberOfTeams, recentlyCreated).
     • page (integer, optional): Page number for pagination. Default 1.
   • Returns: List of competitions with title, ref, category, deadline, and team count.

2. competition_files — List data files for a competition.

   • Inputs:
     • competition (string, required): Competition URL suffix (e.g. titanic).
   • Returns: List of file names and sizes.

3. competition_download — Download competition data files. Returns download URL.

   • Inputs:
     • competition (string, required): Competition URL suffix.
     • file_name (string, optional): Specific file to download. Empty for all files.
   • Returns: Download URL.

4. competition_submit — Submit predictions to a competition.

   • Inputs:
     • competition (string, required): Competition URL suffix.
     • blob_file_tokens (string, required): Blob file token from upload.
     • message (string, required): Submission description message.
   • Returns: Submission result details.

5. competition_submissions — View submission history for a competition.

   • Inputs:
     • competition (string, required): Competition URL suffix.
   • Returns: List of submissions with date, score, status, and description.

6. competition_leaderboard — View competition leaderboard (top 20).

   • Inputs:
     • competition (string, required): Competition URL suffix.
   • Returns: Top 20 team names and scores.

1. datasets_list — Search and list Kaggle datasets.

   • Inputs:
     • search (string, optional): Search term.
     • sort_by (string, optional): Sort order (hottest, votes, updated, active).
     • file_type (string, optional): Filter by file type (csv, json, sqlite, etc).
     • page (integer, optional): Page number. Default 1.
   • Returns: List of datasets with title, ref, size, and download count.

2. dataset_files — List files in a dataset.

   • Inputs:
     • owner (string, required): Dataset owner username.
     • dataset_slug (string, required): Dataset slug name.
   • Returns: List of file names and sizes.

3. dataset_download — Download dataset files. Returns download URL.

   • Inputs:
     • owner (string, required): Dataset owner username.
     • dataset_slug (string, required): Dataset slug name.
     • file_name (string, optional): Specific file. Empty for all.
   • Returns: Download URL.

4. dataset_metadata — Get dataset metadata.

   • Inputs:
     • owner (string, required): Dataset owner username.
     • dataset_slug (string, required): Dataset slug name.
   • Returns: Dataset metadata dictionary.

5. dataset_create — Create a new dataset. Use file_upload first to get file tokens.

   • Inputs:
     • owner (string, required): Owner username.
     • slug (string, required): Dataset slug.
     • title (string, required): Dataset title.
     • file_tokens (string, optional): Comma-separated file tokens from file_upload.
     • license_name (string, optional): License (e.g. CC0-1.0, CC-BY-SA-4.0). Default CC0-1.0.
     • is_private (boolean, optional): Whether dataset is private. Default true.
   • Returns: Creation result details.

6. file_upload — Upload a file to Kaggle and get a token for dataset_create.

   • Inputs:
     • file_name (string, required): File name (e.g. data.csv, config.json).
     • content (string, required): File content as text.
   • Returns: File token string.

1. kernels_list — Search and list Kaggle notebooks/kernels.

   • Inputs:
     • search (string, optional): Search term.
     • competition (string, optional): Filter by competition.
     • dataset (string, optional): Filter by dataset.
     • sort_by (string, optional): Sort order (hotness, commentCount, dateCreated, dateRun, relevance, voteCount).
     • page (integer, optional): Page number. Default 1.
   • Returns: List of kernels with title, ref, votes, and language.

2. kernel_pull — Get a notebook's source code.

   • Inputs:
     • user_name (string, required): Kernel owner username.
     • kernel_slug (string, required): Kernel slug name.
   • Returns: Kernel metadata and source code.

3. kernel_push — Push/save a notebook to Kaggle.

   • Inputs:
     • title (string, required): Notebook title.
     • text (string, required): Notebook source code.
     • language (string, optional): Language (python, r). Default python.
     • kernel_type (string, optional): Type (notebook, script). Default notebook.
     • is_private (boolean, optional): Whether notebook is private. Default true.
   • Returns: Push result details.

1. models_list — Search and list Kaggle models.

   • Inputs:
     • search (string, optional): Search term.
     • owner (string, optional): Filter by owner.
     • sort_by (string, optional): Sort order (hotness, downloadCount, createTime, updateTime).
     • page_size (integer, optional): Number of results per page. Default 20.
   • Returns: List of models with title and ref.

2. model_get — Get detailed information about a specific model.

   • Inputs:
     • owner (string, required): Model owner username.
     • model_slug (string, required): Model slug/name.
   • Returns: Model metadata dictionary.

1. discussions_search — Search Kaggle discussions.

   • Inputs:
     • query (string, required): Search query string.
     • page_size (integer, optional): Number of results (max 50). Default 20.
   • Returns: List of discussions with ID, title, and votes.

2. discussions_list — List discussions for a competition or dataset.

   • Inputs:
     • competition (string, optional): Competition slug to filter.
     • dataset (string, optional): Dataset ref to filter.
     • page_size (integer, optional): Number of results (max 50). Default 20.
   • Returns: List of discussions with ID, title, and votes.

3. discussion_detail — Get discussion content by ID.

   • Inputs:
     • discussion_id (integer, required): Numeric discussion ID.
   • Returns: Discussion title, author, votes, forum, and body content.

4. discussion_comments — Get comments for a discussion.

   • Inputs:
     • discussion_id (integer, required): Numeric discussion ID.
   • Returns: Link to discussion comments page.