DataRobot

Tool to add file(s) to an existing DataRobot files catalog item. Use when you need to upload additional files to an existing catalog entry. The file addition is asynchronous - poll the returned location URL to check status.

Untrusted display-only summary

Tool to apply staged files to a catalog item. Use when you need to finalize and commit files from a stage to a files catalog item.

Untrusted display-only summary

Tool to create a files catalog item in DataRobot from a URL. Use when you need to import files from a publicly accessible URL into the DataRobot catalog. The file is downloaded and processed asynchronously - use GET_CATALOG_ITEM to check the import status.

Untrusted display-only summary

Tool to add file(s) into an existing files catalog item from a URL. Use when you need to add additional files from a URL to an existing DataRobot files catalog item. The file download and processing happens asynchronously - poll the returned location URL to check status.

Untrusted display-only summary

Tool to generate temporary download URLs for files in a catalog item. Use when you need to retrieve data from an unstructured dataset via direct download links. The generated URLs are temporary and valid for the specified duration (default 10 minutes, max 50 minutes).

Untrusted display-only summary

Tool to create an empty stage for File Registry files upload. Use when you need to upload files to DataRobot's File Registry. The stage acts as a temporary upload container before finalizing the file.

Untrusted display-only summary

Tool to stage a file for a batch upload in DataRobot. Use when uploading files to a specific catalog item's stage. Returns the catalog ID and stage ID upon successful upload.

Untrusted display-only summary

Tool to generate temporary download URLs for catalog file versions. Use when you need to retrieve data from a catalog item and version via URLs. The generated URLs expire after the specified duration (default 600 seconds, maximum 3000 seconds).

Untrusted display-only summary

Tool to create a GenAI chat prompt and execute it with an LLM. Use when you want to send a prompt to a DataRobot LLM for completion, optionally leveraging vector database retrieval for RAG workflows.

Untrusted display-only summary

Tool to create a new GenAI comparison chat within a playground for comparing different LLM responses. Use when you need to set up a new comparison chat environment to evaluate multiple model outputs side-by-side.

Untrusted display-only summary

Tool to create and run validation tests for GenAI custom model embeddings. Use when you need to verify that a custom model deployment meets DataRobot's requirements for embedding generation before using it in production.

Untrusted display-only summary

Tool to create a GenAI custom model LLM validation. Use when you need to test and validate LLM deployment behavior. Returns a validation object with status TESTING initially; poll the validation status to check completion.

Untrusted display-only summary

Create a validation for a GenAI custom model deployment's vector database capabilities. Returns immediately with validation ID - validation runs asynchronously and status can be checked using the returned ID. Use when you need to verify that a custom model deployment can properly handle vector database operations for…

Untrusted display-only summary

Tool to create a GenAI custom model version from an LLM blueprint in DataRobot. Use when you need to register a custom model version for deployment or testing. Prerequisites: You must first create an LLM blueprint before calling this endpoint.

Untrusted display-only summary

Tool to create a GenAI evaluation dataset configuration for testing LLM applications. Use when you need to set up evaluation datasets with prompt columns and optional response/tool call columns for assessing GenAI playground outputs. Required for running AI robustness tests on GenAI use cases.

Untrusted display-only summary

Tool to create a new GenAI LLM blueprint for generative AI applications. Use when you need to configure an LLM with specific settings, system prompts, and optional RAG capabilities. Requires a valid playgroundId and llmId. Set llmId to 'custom-model' when using customModelLlmSettings.

Untrusted display-only summary

Tool to create a GenAI LLM blueprint from an existing chat prompt. Use when you have a chat prompt ID and want to convert it into a reusable LLM blueprint for experimentation and deployment.

Untrusted display-only summary

Tool to create a new GenAI LLM Blueprint by copying from an existing LLM Blueprint. Use when you need to duplicate an LLM blueprint configuration with modifications or for version control. The new blueprint inherits all properties from the source blueprint except for the name and description.

Untrusted display-only summary

Tool to create a GenAI LLM test configuration for AI robustness testing. Use when you need to set up automated testing for LLM outputs against metrics like toxicity, jailbreak attempts, or PII leakage.

Untrusted display-only summary

Tool to create a new GenAI LLM test suite for evaluating LLM applications. Use when you need to group LLM test configurations together for AI robustness testing. Requires a valid useCaseId - use DATAROBOT_LIST_USE_CASES to find available use cases.

Untrusted display-only summary

Tool to create a new GenAI playground for experimenting with LLM applications. Use when you need to set up a new playground environment associated with a specific use case.

Untrusted display-only summary

Tool to create OOTB (Out-of-the-Box) metric configurations for a GenAI playground. Use when you need to add performance, safety, and governance metrics to monitor LLM responses in a playground.

Untrusted display-only summary

Tool to create a trace dataset for a GenAI playground. Use when you need to create a dataset for storing playground traces associated with a specific playground ID.

Untrusted display-only summary

Tool to apply guard configurations to a new custom model version in DataRobot. Creates a new version with the specified moderation guards. Use when you need to add toxicity detection, PII filtering, or other content moderation to a custom model.

Untrusted display-only summary

Create a guard configuration to monitor and control AI model behavior during prompt and response stages. Guards can detect issues like PII, toxic content, off-topic responses, or custom validation rules. Requires a templateId (from DATAROBOT_LIST_GUARD_TEMPLATES) and entityId (custom model or playground).

Untrusted display-only summary

Tool to request calculation of a Lift chart insight in DataRobot. Use when you need to analyze model performance across population segments with optional data slicing. The Lift chart calculation is asynchronous. Poll the returned location URL to check status and retrieve results once calculation completes.

Untrusted display-only summary

Request calculation of ROC curve insights for a DataRobot model, custom model, or vector database. The ROC curve is computed asynchronously and can optionally be filtered by a data slice. Use this to evaluate binary classification model performance across different data partitions.

Untrusted display-only summary

Tool to request calculation of SHAP Distributions in DataRobot. Use when you need to analyze the distribution of SHAP values across features with optional data slicing. SHAP (SHapley Additive exPlanations) Distributions help understand how feature values contribute to model predictions across the dataset. The calculat…

Untrusted display-only summary

Tool to request calculation of SHAP Impact insights in DataRobot. Use when you need to understand feature importance based on Shapley values with optional data slicing. SHAP (SHapley Additive exPlanations) Impact provides model-agnostic feature importance by quantifying how much each feature contributes to individual…

Untrusted display-only summary

Tool to request calculation of a SHAP Matrix insight in DataRobot. Use when you need feature importance explanations showing how each feature impacts model predictions, with optional data slicing. The SHAP Matrix calculation is asynchronous. Poll the returned location URL to check status and retrieve results once calc…

Untrusted display-only summary