Skip to main content
Glama
Galaxy-Dawn

kaggle-mcp

by Galaxy-Dawn
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Remote

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
KAGGLE_API_TOKENNoYour Kaggle API token. Alternative: place kaggle.json file in ~/.kaggle/ with username and key.

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
benchmark_leaderboardB

Get the leaderboard for a Kaggle benchmark.

    Args:
        owner_slug: Benchmark owner username or organization slug.
        benchmark_slug: Benchmark slug name.
        version_number: Specific version number (0 for latest).
competitions_listA

Search and list Kaggle competitions.

    Args:
        search: Search term to filter competitions.
        category: Filter by category (e.g. featured, research, playground).
        sort_by: Sort order (latestDeadline, numberOfTeams, recentlyCreated).
        page: Page number for pagination.
competition_filesA

List data files for a competition.

    Args:
        competition: Competition URL suffix (e.g. 'titanic').
competition_downloadB

Download competition data files. Returns download URL.

    Args:
        competition: Competition URL suffix (e.g. 'titanic').
        file_name: Specific file to download. Empty for all files.
competition_submitB

Submit predictions to a competition.

    Args:
        competition: Competition URL suffix.
        blob_file_tokens: Blob file token from upload.
        message: Submission description message.
competition_submissionsA

View submission history for a competition.

    Args:
        competition: Competition URL suffix.
competition_leaderboardC

View competition leaderboard (top 20).

    Args:
        competition: Competition URL suffix.
competition_getB

Get detailed competition info.

    Args:
        competition: Competition URL suffix (e.g. 'titanic').
competition_data_summaryA

Get data files summary for a competition.

    Args:
        competition: Competition URL suffix (e.g. 'titanic').
competition_get_submissionB

Get details for a single submission.

    Args:
        competition: Competition URL suffix (unused in routing, kept for context).
        submission_id: Numeric submission ID.
competition_leaderboard_downloadA

Download the full competition leaderboard. Returns download URL.

    Args:
        competition: Competition URL suffix (e.g. 'titanic').
datasets_listB

Search and list Kaggle datasets.

    Args:
        search: Search term.
        sort_by: Sort order (hottest, votes, updated, active).
        file_type: Filter by file type (csv, json, sqlite, etc).
        page: Page number.
dataset_filesB

List files in a dataset.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
dataset_downloadB

Download dataset files. Returns download URL.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
        file_name: Specific file. Empty for all.
dataset_metadataC

Get dataset metadata.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
dataset_createA

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

    Args:
        owner: Owner username.
        slug: Dataset slug.
        title: Dataset title.
        file_tokens: Comma-separated file tokens from file_upload.
        license_name: License (e.g. CC0-1.0, CC-BY-SA-4.0).
        is_private: Whether dataset is private.
file_uploadA

Upload a file to Kaggle and get a token for dataset_create.

    Args:
        file_name: File name (e.g. 'data.csv', 'config.json').
        content: File content as text.
dataset_getA

Get full dataset info.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
dataset_create_versionA

Create a new version of an existing dataset.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
        version_notes: Notes describing this version.
        file_tokens: Comma-separated file tokens from file_upload.
dataset_update_metadataB

Update dataset metadata (title, description, license).

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
        title: New title (leave empty to keep current).
        description: New description (leave empty to keep current).
        license_name: New license name (leave empty to keep current).
dataset_deleteC

Delete a dataset.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
dataset_download_fileB

Download a single file from a dataset. Returns download URL.

    Args:
        owner: Dataset owner username.
        dataset_slug: Dataset slug name.
        file_name: Name of the specific file to download.
kernels_listB

Search and list Kaggle notebooks/kernels.

    Args:
        search: Search term.
        competition: Filter by competition.
        dataset: Filter by dataset.
        sort_by: Sort order (hotness, commentCount, dateCreated, dateRun, relevance, voteCount).
        page: Page number.
kernel_pullC

Get a notebook's source code.

    Args:
        user_name: Kernel owner username.
        kernel_slug: Kernel slug name.
kernel_pushC

Push/save a notebook to Kaggle.

    Args:
        title: Notebook title.
        text: Notebook source code.
        language: Language (python, r).
        kernel_type: Type (notebook, script).
        is_private: Whether notebook is private.
kernel_outputC

Download competition data files. Returns download URL.

    Args:
        user_name: Kernel owner username.
        kernel_slug: Kernel slug name.
kernel_session_createC

Create an interactive kernel session.

    Args:
        user_name: Kernel owner username.
        kernel_slug: Kernel slug name.
kernel_session_statusC

Get status of a kernel session.

    Args:
        user_name: Kernel owner username.
        kernel_slug: Kernel slug name.
kernel_session_outputB

List output files from a kernel session.

    Args:
        user_name: Kernel owner username.
        kernel_slug: Kernel slug name.
kernel_session_cancelC

Cancel a running kernel session.

    Args:
        user_name: Kernel owner username.
        kernel_slug: Kernel slug name.
competition_top_kernelsA

List top public kernels/notebooks for a competition, sorted by public score.

    Note: Kaggle API does not return the score value for active competitions,
    but DOES sort by score. Scores shown are extracted from notebook titles
    (e.g. '[44/50]', '[0.371]') where authors include them.

    Args:
        competition: Competition URL suffix (e.g. 'titanic').
        sort_by: Sort order — scoreDescending (default), scoreAscending,
                 voteCount, hotness, dateCreated, dateRun, commentCount.
        page_size: Number of results (max 100).
models_listA

Search and list Kaggle models.

    Args:
        search: Search term.
        owner: Filter by owner.
        sort_by: Sort order (hotness, downloadCount, createTime, updateTime).
        page_size: Number of results per page.
model_getC

Get detailed information about a specific model.

    Args:
        owner: Model owner username.
        model_slug: Model slug/name.
model_createB

Create a new Kaggle model.

    Args:
        owner: Owner username.
        slug: Model slug (URL-friendly identifier).
        title: Model title.
        subtitle: Optional subtitle.
        description: Optional description.
        is_private: Whether the model is private (default True).
model_updateA

Update an existing Kaggle model.

    Args:
        owner: Owner username.
        model_slug: Model slug to update.
        title: New title (leave empty to keep existing).
        subtitle: New subtitle (leave empty to keep existing).
        description: New description (leave empty to keep existing).
        is_private: New privacy setting (None to keep existing).
model_deleteB

Delete a Kaggle model.

    Args:
        owner: Owner username.
        model_slug: Model slug to delete.
model_instances_listA

List all instances of a Kaggle model.

    Args:
        owner: Model owner username.
        model_slug: Model slug/name.
model_instance_getB

Get details of a specific model instance.

    Args:
        owner: Model owner username.
        model_slug: Model slug/name.
        framework: Framework name (e.g. 'tensorflow2', 'pytorch', 'jax').
        instance_slug: Instance slug identifier.
model_instance_createB

Create a new instance for a Kaggle model.

    Args:
        owner: Model owner username.
        model_slug: Model slug/name.
        framework: Framework name (e.g. 'tensorflow2', 'pytorch', 'jax').
        instance_slug: Slug for the new instance.
        overview: Optional overview text.
        usage: Optional usage instructions.
        license_name: License name (default 'Apache 2.0').
        is_private: Whether the instance is private (default True).
model_instance_versionsB

List all versions of a model instance.

    Args:
        owner: Model owner username.
        model_slug: Model slug/name.
        framework: Framework name (e.g. 'tensorflow2', 'pytorch', 'jax').
        instance_slug: Instance slug identifier.
model_instance_version_createC

Create a new version for a model instance.

    Args:
        owner: Model owner username.
        model_slug: Model slug/name.
        framework: Framework name (e.g. 'tensorflow2', 'pytorch', 'jax').
        instance_slug: Instance slug identifier.
        version_notes: Optional notes for this version.
discussions_searchA

Search Kaggle discussions.

    Args:
        query: Search query string.
        sort_by: Sort order: hotness, votes, comments, created, updated.
        source_type: Filter by source: competition, dataset, kernel, site_forum,
            competition_solution, model, write_up, learn_track, benchmark, benchmark_task.
        page_size: Number of results (max 50).
discussions_listA

List discussions for a competition or dataset.

    Args:
        competition: Competition slug to filter.
        dataset: Dataset ref to filter.
        page_size: Number of results (max 50).
        since_hours: If > 0, only return discussions within the last N hours.
                     Fetches up to 100 results and filters client-side.
        new_only: If True, filter by createTime (newly posted threads only).
                  If False (default), filter by updateTime (includes threads with new replies).
discussion_detailA

Get discussion content by ID.

    Args:
        discussion_id: Numeric discussion ID.
        competition: Competition slug to narrow the search (recommended for accuracy).
discussion_commentsB

Get comments for a discussion.

    Args:
        discussion_id: Numeric discussion ID.
        page_size: Number of comments to return (max 50).
discussion_comments_searchB

Search comments across all Kaggle discussions.

    Args:
        query: Search query string.
        page_size: Number of results (max 50).
discussions_by_sourceB

Browse discussions by source type.

    Args:
        source_type: One of: competition, dataset, kernel, site_forum,
            competition_solution, model, write_up, learn_track, benchmark, benchmark_task.
        query: Optional search query to filter results.
        sort_by: Sort order: hotness, votes, comments, created, updated.
        page_size: Number of results (max 50).
discussions_solutionsB

Browse competition solution write-ups.

    Args:
        competition: Optional competition slug to filter solutions.
        sort_by: Sort order: hotness, votes, comments, created, updated.
        page_size: Number of results (max 50).
discussions_writeupsA

Browse Kaggle write-ups by type.

    Args:
        write_up_type: One of: knowledge, competition_solution, hackathon,
            personal_project, forum_topic, blog.
        query: Optional search query to filter results.
        page_size: Number of results (max 50).
discussions_trendingB

Browse trending discussions sorted by hotness.

    Args:
        source_type: Optional filter: competition, dataset, kernel, site_forum,
            competition_solution, model, write_up, learn_track, benchmark, benchmark_task.
        page_size: Number of results (max 50).
discussions_myB

List the current user's discussions.

    Args:
        page_size: Number of results (max 50).

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.4/5.0

Scored across 51 tools

Disambiguation5/5

Every tool has a clear resource prefix (competition_, dataset_, kernel_, model_, discussion_) followed by a distinct action verb. Overlaps are minimal and resolved by descriptions (e.g., dataset_download vs. dataset_download_file). The only minor issue is kernel_output's description being a copy-paste error, but tool names are unambiguous.

Naming Consistency5/5

All tool names follow a consistent snake_case pattern of resource_prefix + action verb (e.g., competition_get, dataset_create, discussion_comments). No mixing of styles across the entire set.

Tool Count3/5

With 51 tools, this is on the high side for a single MCP server. While each tool serves a specific purpose and the Kaggle API is broad, the number feels heavy, especially with 12+ discussion-related tools. It's borderline but still manageable for an agent.

Completeness5/5

The tool surface covers nearly all key Kaggle resources: competitions (list, get, submit, leaderboard, data), datasets (full CRUD, versions, downloads), kernels (push, pull, sessions), models (create, delete, instances, versions), discussions (search, list, comments), and file upload. Missing actions (e.g., delete submission/kernel) are minor and uncommon. Very comprehensive.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources