Create eval run records from incoming webhook reports

Your AI eval summaries stay in chat without structured tracking, so engineers can't audit runs or compare results. You capture run records with chat context for quick review and same-day triage.