Class Client

Create a new configuration

Create a new LLM or pipeline configuration with provider, parameters, and environment settings.

Delete a configuration

Permanently delete a configuration by its unique identifier.

Retrieve a list of configurations

List configurations with optional filtering by name, environment, and tags.

Update an existing configuration

Update an existing configuration's name, provider, parameters, environment, or tags.

Create a new datapoint

Create a single datapoint with inputs, history, ground truth, and metadata.

Create multiple datapoints in batch

Create multiple datapoints from events using field mappings and optional filters.

Delete a specific datapoint

Permanently delete a datapoint by its unique identifier.

Retrieve a specific datapoint

Get a single datapoint by its unique identifier.

Retrieve a list of datapoints

Retrieve datapoints, optionally filtered by a list of datapoint IDs or dataset name.

Update a specific datapoint

Update fields on an existing datapoint. Only the provided fields are modified.

Add datapoints to a dataset

Add new datapoints to an existing dataset. Provide raw data objects and a field mapping that specifies which fields map to inputs, ground truth, and history.

Create a dataset

Create a new dataset with an optional name, description, and initial set of datapoint IDs.

Delete a dataset

Permanently delete a dataset by its unique identifier.

Get datasets

Retrieve datasets, optionally filtered by dataset ID or name.

Remove a datapoint from a dataset

Remove a specific datapoint from a dataset. The datapoint itself is not deleted, only dereferenced from the dataset.

Update a dataset

Update a dataset's name, description, or list of datapoint IDs.

Create a new event

Create a new event (span) within a session trace. The request body wraps the event object under the event key.

Required properties within the event object:

• event_type (string) — Must be one of: chain, model, tool, session.
• inputs (object) — Input data for the event. Auto-generated properties (provided by the server when omitted):
• event_id (string, UUID) — Unique identifier for the event.
• session_id (string, UUID) — Session/trace identifier.
• parent_id (string, UUID) — Parent event ID. Defaults to session_id. Optional properties with defaults:
• event_name (string) — Name of the event. Defaults to "unknown".
• source (string) — Source of the event (e.g. sdk-python). Defaults to "unknown". Optional properties:
• config (object) — Configuration data (e.g. model parameters, prompt templates).
• outputs (object) — Output data from the event.
• error (string or null) — Error message if the event failed.
• children_ids (array of strings) — IDs of child events.
• duration (number) — Duration of the event in milliseconds.
• start_time (number) — Unix timestamp in milliseconds for event start.
• end_time (number) — Unix timestamp in milliseconds for event end.
• metadata (object) — Additional metadata (e.g. token counts, cost).
• metrics (object) — Custom metrics.
• feedback (object) — Feedback data (e.g. ratings, ground truth).
• user_properties (object) — User properties associated with the event.

Create a batch of events

Create multiple events in a single request. When single_session is true, all events share the same session. Please refer to our instrumentation guide for detailed information.

Create a new model event

Create a model event. The event_type is automatically set to 'model'. Please refer to our instrumentation guide for detailed information.

Create a batch of model events

Create multiple model events in a single request. The event_type is automatically set to 'model' for all events. When single_session is true, all events share the same session. Please refer to our instrumentation guide for detailed information.

Retrieve events based on filters

Export events via POST with filtering, projections, and pagination. This is the primary method for retrieving events from HoneyHive.

Update an event

Update fields on an existing event. Only the provided fields are modified; omitted fields are left unchanged. The event_id field is required to identify the event to update.

Compare events between two experiment runs

Retrieve and compare events between two experiment runs for detailed analysis

Retrieve experiment comparison

Compare metrics and results between two experiment runs

Create a new evaluation run

Create a new experiment run to track an evaluation against a dataset.

Delete an evaluation run

Permanently delete an experiment run by its run ID.

Retrieve experiment result

Compute evaluation summary for an experiment run including pass/fail status, metrics, and datapoints

Get details of an evaluation run

Retrieve the full details of a single experiment run by its run ID.

Get event metrics for an experiment run

Retrieve event metrics from ClickHouse for a specific experiment run

Get experiment runs schema

Retrieve the schema and metadata for experiment runs

Get a list of evaluation runs

List experiment runs with optional filtering by dataset, status, name, date range, and specific run IDs. Results are paginated and sortable.

Update an evaluation run

Update fields on an existing experiment run such as name, status, metadata, or results.

Create a new metric

Add a new metric

Delete a metric

Remove a metric

Get all metrics

Retrieve a list of all metrics

Run a metric evaluation

Execute a metric on a specific event

Update an existing metric

Edit a metric

Add traces to a session

Add trace events to an existing session. The field is named logs for legacy compatibility with the Go ingestion handler.

Start a new session

Start a new session. The request body wraps the session event object under the session key, matching the pattern used by POST /events.