2026-07-27T00:14:53.209Z
Site Suitability (clean v6 fixture)
Quality score100%48 of 48 tasks completed · whole number, no decimals
- Source
- GeoWorkBench v0.1 executor · E2B sandbox
- Test
- site-suitability-v6 · 48 accepted of 48
- Output
- 48 accepted · 0 rejected
- Cost
- 0.05023 USD total
- Time
- Deterministic sandbox execution
- Failure
- 0 rejected by verifier
- Limit
- Reporting a count without the selected point features, including an outside or low-score site, must fail.
- Verifier
- site-suitability — deterministic verification first, LLM judge as side metric only
- Evidence
- Sealed v0.1 executor with four versioned deterministic tasks. 96 trials recorded: 84 accepted, 12 rejected.