Skip to main content
The Runs API exposes workflow execution state: run history, step-by-step traces with tool telemetry, produced artifacts, and the operator approval gate. It backs the Run Inspector in the Dashboard, Auditing, and Command Center pages. All endpoints are scoped to the authenticated workspace. Requests without a workspace return HTTP 401.

List runs

Returns the 50 most recent workflow runs for the workspace, newest first.
Run status values include pending, running, paused_for_approval, completed, and failed.

Get run details

Returns the full execution record for one run:
  • run: the run row, including status, trigger source, and initial context.
  • workflow: metadata for the workflow the run belongs to.
  • steps: every run step in execution order, enriched with the step title, step type, action prompt, assigned agent name and role, the tools executed, and the count of artifacts each step created.
  • artifacts: all artifacts the run produced.
  • totalToolsCount, totalCost, totalLatencyMs: aggregate telemetry across all steps.
Returns HTTP 404 if the run does not exist in the workspace.

Approve a paused run

Resumes a run that is waiting at an approval gate. The run must be in status paused_for_approval; approving sets it back to pending and immediately resumes queue processing.
Returns HTTP 400 with "Run is not pending approval" if the run is in any other status, and 404 if the run is not found.

Reject a paused run

Ends a run that is waiting at an approval gate. The run must be in status paused_for_approval; rejecting marks it failed with the error message Rejected by user and records the completion time.
Returns HTTP 400 if the run is not pending approval, and 404 if the run is not found.

Get run artifacts

Returns all artifacts produced by one run, oldest first:
For artifact fields, downloads, and the quality loop, see the Artifacts API.