Browse standalone agent answers and their response scoring. Recompute the response + meta score against the current catalog rubrics with any judge model โ this never touches set-level runs.
Stacktrace: