eval results.md

Eval results

A result is the completed outcome of an eval execution. Run IDs and result IDs are separate. The result includes the judge outcome and supporting details.

Commands

Command Description
lightsage eval-results list List completed eval results.
lightsage eval-results retrieve --result-id <result_id> Retrieve one result with details.

List eval results

lightsage eval-results list --output-format json

Filter by time window:

lightsage eval-results list --since 7d --limit 50

The --since flag accepts a relative window such as 7d or an ISO timestamp.

Retrieve one result

lightsage eval-results retrieve --result-id result_123

The response includes the judge outcome, output, and diagnostic details.