About This Launch
Eval-only JSONL + promptfoo harness for appliance-repair answering agents. EMERGENCY/URGENT/ROUTINE/ESTIMATE plus FAIL-bait. Unzip, swap provider, read the score table. Not a phone bot.
Why I Built This
I built this to make appliance-repair answering agents safer and easier to evaluate before they reach customers. It packages realistic triage cases, clear labels, and repeatable scoring so teams can catch bad guidance quickly.
Launch Details
CategoryAI / ML
StageLive
ListedSeptember 7, 2026