Skip to main content
GET
Read one run and each case execution or evaluation result. Item status is queued, running, completed or failed. Completed results include verdict, criteriaResults, replayInteractionId, environment, messages and Action diagnostics. A failed execution may include diagnostic result data plus error. Previous comparisons use previousResult, previousCreatedAt and sameSetup. Private execution-artifact snapshots are not returned.

Headers

string
required
Bearer token containing your workspace API key. The key needs read scope.

Path parameters

string
required
Run UUID.

Response

Returns the run and items. The example intentionally has an empty array of collection origins and uses recorded conversation messages without Action calls. Queued items return result: null until a result is available.

Failure analysis

A completed result with a failing required check can include items[].result.failureAnalysis. It is optional: older runs and passing results may omit it. Analysis does not replace the criterion verdict or change an execution error into an evaluated result.
string
suggested or inconclusive. A suggestion is not a confirmed cause. Insufficient evidence or an unavailable analysis produces an inconclusive result while preserving the original verdict.
string
Summary of the failed check and observed behavior.
string
Suggested issue, or Needs investigation when evidence is insufficient.
string
Explanation of the suggestion or why the analysis is inconclusive.
array
References with id, kind, and label. Kinds are message, action, rule, article, behaviour, context, or criterion. References may include eventId, a zero-based actionIndex, versionId, or version. The array can be empty when analysis is unavailable; it is not the private execution-artifact snapshot.