Filter to a single workforce component by its numeric app id (the value the SDK reads from TIMBAL_APP_ID, also returned as component.id on each run). Combines with component_uid: when both are set, rows must match both.

Filter to a single workforce component by its manifest uid (the _id field in timbal.yaml, also returned as component.uid on each run). Display name is deliberately not accepted as a filter — it's user-editable and not guaranteed unique within a project.

Optional trigger filter. Accepts:

• latest — runs from the env's most recent deploy batch.
• a UUID — runs from that deploy batch.

Filter by what started the run: deploy (gate), manual (POST .../evals/runs), or push (deploy_config.evals_on_push CI runs). Orthogonal to trigger (a deploy-batch selector).

Filter by run status: running, passed, failed, or error.

Include the captured CLI stdout/stderr in each run. Off by default — logs can be hundreds of KB per run.

Include the structured per-eval results document in each run. Off by default for the same size reason as include_logs; the single-run GET always returns it.

Max rows returned, newest first. Defaults to 50, capped at 200.