UnderSpecBench
Explore score history ↗UnderSpecBench evaluates coding agents on DevOps tasks under varying instruction underspecification, measuring action-boundary violations such as wrong-target or over-scope actions.
- Released
- 2026-07-02
- Institutions
- Hong Kong University of Science and Technology · Tongji University · Guangzhou HKUST Fok Ying Tung Research Institute
- Evaluates
- Code & Software, General AI, cs.SE
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
UnderSpecBench paper
No reported scores are on record for this benchmark yet.