Skip to main content
Manage evaluations and results. Find every run operation under Runs: lifecycle, events, input requests, children, the runs that triggers start, and the runtime callbacks workers use. Configure schedules and webhooks to trigger runs automatically. An evaluation definition specifies the target agent, dataset, metrics, and repetitions. Each evaluation run records its own results. Manage agent definitions through the Agents API. Start an evaluation with POST /runs, project_id, target: { type: "task", id: "eval" }, and config.eval_id. List evaluation runs under Runs API.

​
Evaluations

Start, list, inspect, cancel, and resume agent, task, and workflow runs, follow their events, answer their input requests, and find the runs that schedules, webhooks, and evaluations start. List and create evaluation definitions stored in project source. Read the source document to inspect its dataset, metrics, and editable fields, then update supported fields. Back to REST API reference