Models API

Kaggle-MCP provides tools to search, download, create, and manage Kaggle models directly through Claude.

Available Tools

models_list

List available Kaggle models with filtering options.

models_list(search="", sort_by="hotness", owner="", page_size=20, page_token="")

Parameters:

• search: Term(s) to search for
• sort_by: Sort models by (hotness, votes, updated, active)
• owner: Display models by a specific user or organization
• page_size: Number of items per page
• page_token: Page token for pagination

Returns: JSON string with model details

model_get

Get details of a specific model.

model_get(model)

Parameters:

• model: Model identifier in format <owner>/<model-name>

Returns: JSON string with model details

model_initialize_metadata

Initialize metadata file for model creation.

model_initialize_metadata(path=".")

Parameters:

• path: Directory where metadata file will be created

Returns: Success message or error details

model_create_new

Create a new model on Kaggle.

model_create_new(folder_path)

Parameters:

• folder_path: Path to the folder containing model files and metadata

Returns: Success message or error details

model_update

Update an existing model.

model_update(model, folder_path)

Parameters:

• model: Model identifier in format <owner>/<model-name>
• folder_path: Path to the folder containing model metadata

Returns: Success message or error details

model_delete

Delete a model.

model_delete(model)

Parameters:

• model: Model identifier in format <owner>/<model-name>

Returns: Success message or error details

Model Instances

Model instances represent specific implementations of a model.

model_instance_get

Get details of a specific model instance.

model_instance_get(model_instance)

Parameters:

• model_instance: Model instance identifier in format <owner>/<model-name>/<framework>/<instance-slug>

Returns: JSON string with model instance details

model_instance_initialize_metadata

Initialize metadata file for model instance creation.

model_instance_initialize_metadata(path=".")

Parameters:

• path: Directory where metadata file will be created

Returns: Success message or error details

model_instance_create

Create a new model instance.

model_instance_create(model, folder_path)

Parameters:

• model: Model identifier in format <owner>/<model-name>
• folder_path: Path to the folder containing model instance metadata

Returns: Success message or error details

model_instance_update

Update an existing model instance.

model_instance_update(model_instance, folder_path)

Parameters:

• model_instance: Model instance identifier in format <owner>/<model-name>/<framework>/<instance-slug>
• folder_path: Path to the folder containing model instance metadata

Returns: Success message or error details

model_instance_delete

Delete a model instance.

model_instance_delete(model_instance)

Parameters:

• model_instance: Model instance identifier in format <owner>/<model-name>/<framework>/<instance-slug>

Returns: Success message or error details

model_instance_list_files

List files in a model instance for the current version.

model_instance_list_files(model_instance)

Parameters:

• model_instance: Model instance identifier in format <owner>/<model-name>/<framework>/<instance-slug>

Returns: JSON string with file details

Model Instance Versions

You can manage different versions of a model instance.

model_instance_version_create

Create a new model instance version.

model_instance_version_create(model_instance, folder_path, version_notes)

Parameters:

• model_instance: Model instance identifier in format <owner>/<model-name>/<framework>/<instance-slug>
• folder_path: Path to the folder containing files to upload
• version_notes: Notes describing the new version

Returns: Success message or error details

model_instance_version_download

Download model instance version files.

model_instance_version_download(model_instance_version, path="")

Parameters:

• model_instance_version: Model instance version identifier in format <owner>/<model-name>/<framework>/<instance-slug>/<version-number>
• path: Folder where file(s) will be downloaded (defaults to a temp directory)

Returns: Success message or error details

model_instance_version_delete

Delete a model instance version.

model_instance_version_delete(model_instance_version)

Parameters:

• model_instance_version: Model instance version identifier in format <owner>/<model-name>/<framework>/<instance-slug>/<version-number>

Returns: Success message or error details

model_instance_version_list_files

List files in a specific model instance version.

model_instance_version_list_files(model_instance_version)

Parameters:

• model_instance_version: Model instance version identifier in format <owner>/<model-name>/<framework>/<instance-slug>/<version-number>

Returns: JSON string with file details

Examples

Here are some examples of how to use the model tools with Claude:

# Search for machine learning models
models_list(search="machine learning")

# Get details about a specific model
model_get("username/model-name")

# Create a new model instance
model_instance_initialize_metadata("/path/to/project")
model_instance_create("username/model-name", "/path/to/project")

# Download a specific model instance version
model_instance_version_download("username/model-name/framework/instance-slug/1", "/path/to/download")