SDK Reference¶

Experimental SDK

The dataset management SDK is under logfire.experimental.api_client. The API may change in future releases.

For usage examples, see the SDK Guide.

LogfireAPIClient¶

Method	Description
`list_datasets()`	List all datasets in the project.
`get_dataset(id_or_name, input_type, output_type, metadata_type, *, include_cases)`	Get a dataset, optionally as a typed `pydantic_evals.Dataset`. Pass `include_cases=False` to get metadata only.
`create_dataset(name, *, input_type, output_type, metadata_type, description, guidance)`	Create a new dataset. Types are converted to JSON schemas automatically.
`update_dataset(id_or_name, *, name, input_type, output_type, metadata_type, description)`	Update a dataset's metadata or schemas.
`delete_dataset(id_or_name)`	Delete a dataset and all its cases.
`list_cases(dataset_id_or_name, *, tags)`	List all cases in a dataset. Optionally filter by tags.
`get_case(dataset_id_or_name, case_id)`	Get a specific case.
`add_cases(dataset_id_or_name, cases, *, tags, on_conflict)`	Add cases to a dataset. Accepts `pydantic_evals.Case` objects or plain dicts. Uses upsert by default (`on_conflict='update'`): cases with matching names are updated. Set `on_conflict='error'` to fail on conflicts.
`update_case(dataset_id_or_name, case_id, *, name, inputs, expected_output, metadata, evaluators, tags)`	Update an existing case.
`delete_case(dataset_id_or_name, case_id)`	Delete a case.

An async version, AsyncLogfireAPIClient, provides the same methods as async coroutines.

Exception	Description
`DatasetNotFoundError`	Raised when a dataset lookup by ID or name finds no match.
`CaseNotFoundError`	Raised when a case lookup finds no match.
`DatasetApiError`	Raised for other API errors. Contains `status_code` and `detail`.