Prompt
Scrape the pricing table into JSON `{plan, monthlyUsd}`. Graded pass@3 across three trials; JSON must match gold each time despite random DOM attribute noise.
Success criteria
- JSON matches gold each trial
- pass@3 = 1.0
- No invented plans
Scoring hints
- reliabilityConsistency
- Prefer semantic selectors over brittle XPath
Budget 700s
Tools required · web yes · code no
Tags extract, multi-trial, pass@3, dom
Primary type Browser / computer-use agent. Gold is not in this payload. API