Databricks

Tool to log dataset inputs to an MLflow run for tracking data sources used during model development. Use when you need to track metadata about datasets used in ML experiment runs, including information about the dataset source, schema, and tags. Enables logging of dataset inputs to a run, allowing you to track data so…

Untrusted display-only summary

Tool to log parameters for a logged model in MLflow. Use when you need to attach hyperparameters or metadata to a LoggedModel object. A param can be logged only once for a logged model, and attempting to overwrite an existing param will result in an error. Available in MLflow 2.8+.

Untrusted display-only summary

Tool to log a metric for an MLflow run with timestamp. Use when you need to record ML model performance metrics like accuracy, loss, or custom evaluation metrics. Metrics can be logged multiple times with different timestamps and values are never overwritten - each log appends to the metric history for that key.

Untrusted display-only summary

Tool to log a model artifact for an MLflow run (Experimental API). Use when you need to record model metadata including artifact paths, flavors, and versioning information for a training run. The model_json parameter should contain a complete MLmodel specification in JSON string format.

Untrusted display-only summary

Tool to log dataset outputs from an MLflow run for tracking data generated during model development. Use when you need to track metadata about datasets produced by ML experiment runs, including information about predictions, model outputs, or generated data. Enables logging of dataset outputs to a run, allowing you to…

Untrusted display-only summary

Tool to log a parameter for an MLflow run as a key-value pair. Use when you need to record hyperparameters or constant values for ML model training or ETL pipelines. Parameters can only be logged once per run and cannot be changed after logging. Logging identical parameters is idempotent.

Untrusted display-only summary

Tool to restore a deleted MLflow experiment and its associated metadata, runs, metrics, params, and tags. Use when you need to recover a previously deleted experiment from Databricks. If the experiment uses FileStore, underlying artifacts are also restored.

Untrusted display-only summary

Tool to restore a deleted MLflow run and its associated metadata, runs, metrics, params, and tags. Use when you need to recover a previously deleted run from Databricks ML experiments. The operation cannot restore runs that were permanently deleted.

Untrusted display-only summary

Tool to bulk restore runs in an ML experiment that were deleted at or after a specified timestamp. Use when you need to recover multiple deleted experiment runs. Only runs deleted at or after the specified timestamp are restored. The maximum number of runs that can be restored in one operation is 10000.

Untrusted display-only summary

Tool to search for logged models in MLflow experiments based on various criteria. Use when you need to find models that match specific metrics, parameters, tags, or attributes using SQL-like filter expressions. Supports pagination, ordering results, and filtering by datasets.

Untrusted display-only summary

Tool to set a tag on an MLflow experiment. Use when you need to add or update experiment metadata. Experiment tags are metadata that can be updated at any time.

Untrusted display-only summary

Tool to set tags on a logged model in MLflow. Use when you need to add or update metadata tags on a LoggedModel object for organization and tracking. Tags are key-value pairs that can be used to search and filter logged models. Part of MLflow 3's logged model management capabilities.

Untrusted display-only summary

Tool to set permissions for an MLflow experiment, replacing all existing permissions. Use when you need to configure access control for an experiment. This operation replaces ALL existing permissions; for incremental updates, use the update permissions endpoint instead.

Untrusted display-only summary

Tool to set a tag on an MLflow run. Use when you need to add custom metadata to runs for filtering, searching, and organizing experiments. Tags with the same key can be overwritten by successive writes. Logging the same tag (key, value) is idempotent.

Untrusted display-only summary

Tool to update MLflow experiment metadata, primarily for renaming experiments. Use when you need to rename an existing experiment. The new experiment name must be unique across all experiments in the workspace.

Untrusted display-only summary

Tool to incrementally update permissions for an MLflow experiment. Use when you need to modify specific permissions without replacing the entire permission set. This PATCH operation updates only the specified permissions, preserving existing permissions not included in the request.

Untrusted display-only summary

Tool to update MLflow run metadata including status, end time, and run name. Use when a run's status changes outside normal execution flow or when you need to rename a run. This endpoint allows you to modify a run's metadata after it has been created.

Untrusted display-only summary

Tool to delete a Kafka configuration from ML Feature Engineering. Use when you need to remove Kafka streaming source configurations. The deletion is permanent and cannot be undone. Kafka configurations define how features are streamed from Kafka sources.

Untrusted display-only summary

Tool to create a Databricks Online Feature Store for real-time feature serving. Use when you need to establish serverless infrastructure for low-latency access to feature data at scale. Requires Databricks Runtime 16.4 LTS ML or above, or serverless compute.

Untrusted display-only summary

Tool to delete an online store from ML Feature Store. Use when you need to remove online stores that provide low-latency feature serving infrastructure. The deletion is permanent and cannot be undone. Online stores are used for real-time feature retrieval in production ML serving.

Untrusted display-only summary

Tool to create a new AutoML forecasting experiment for time series prediction. Use when you need to automatically train and optimize forecasting models on time series data. The experiment will train multiple models and select the best one based on the primary metric.

Untrusted display-only summary

Delete a metadata tag from a specific feature column in a Databricks ML Feature Store table. This operation removes the tag association from the feature but does not affect the actual feature data. The operation is idempotent - it succeeds even if the tag doesn't exist, making it safe to call multiple times. Use this…

Untrusted display-only summary

Tool to retrieve a specific tag from a feature in a feature table in ML Feature Store. Use when you need to get metadata tag details from specific features. This operation returns the tag name and value associated with the feature.

Untrusted display-only summary

Tool to set or update a tag on a feature in a feature table in ML Feature Store. Use when you need to add or modify metadata tags on specific features. If the tag already exists, it will be updated with the new value. If the tag doesn't exist, it will be created automatically. This operation is idempotent and can be u…

Untrusted display-only summary

Retrieves the list of available permission levels that can be assigned to users or groups for a Databricks ML registered model. This endpoint returns metadata about what permission levels are available (e.g., CAN_READ, CAN_EDIT, CAN_MANAGE, CAN_MANAGE_PRODUCTION_VERSIONS, CAN_MANAGE_STAGING_VERSIONS) with descriptions…

Untrusted display-only summary

Tool to delete an OAuth secret from a service principal at the account level. Use when you need to revoke OAuth credentials for service principal authentication. Once deleted, applications or scripts using tokens generated from that secret will no longer be able to access Databricks APIs.

Untrusted display-only summary

Tool to create an OAuth secret for service principal authentication. Use when you need to obtain OAuth access tokens for accessing Databricks Accounts and Workspace APIs. A service principal can have up to five OAuth secrets, each valid for up to two years (730 days). The secret value is only shown once upon creation.

Untrusted display-only summary

Tool to delete a Databricks Delta Live Tables pipeline permanently and stop any active updates. Use when you need to remove a pipeline completely. If the pipeline publishes to Unity Catalog, deletion will cascade to all pipeline tables. This action cannot be easily undone without Databricks support assistance.

Untrusted display-only summary

Tool to retrieve permissions for a Databricks Delta Live Tables pipeline. Use when you need to check who has access to a pipeline and their permission levels. Returns the complete permissions information including access control lists with user, group, and service principal permissions.

Untrusted display-only summary

Tool to retrieve available permission levels for a Databricks Delta Live Tables pipeline. Use when you need to understand what permission levels can be assigned to users or groups for a specific pipeline. Returns permission levels like CAN_VIEW, CAN_RUN, CAN_MANAGE, and IS_OWNER with their descriptions.

Untrusted display-only summary