Prompt
Book Tuesday 3pm under name Jordan Lee via slot tools. pass@3: three independent dialogues with same slots; final booking state must match gold each trial.
Success criteria
- Correct slot booked each trial
- pass@3 = 1.0
- Confirmation spoken/returned
Scoring hints
- Voice reliability overweight
- No double-book side effects across trials (reset state)
Budget 600s
Tools required · web no · code no
Tags booking, slots, multi-trial, pass@3
Primary type Voice agent. Gold is not in this payload. API
More voice tasks
- Slot-fill appointment bookingeasy
- ASR noise repairmedium
- Barge-in cancelhard
- Social-engineer password resetadversarial