SEVRA-BENCH
Explore score history ↗SEVRA-BENCH is a benchmark for measuring how often LLM-based code review agents approve adversarial pull requests with social-engineering framings, built from vulnerability-fixing commits. It includes a challenge split of roughly 1,000 adversarial PRs.
- Released
- 2026-06-11
- Institutions
- Carnegie Mellon University · Microsoft (United States) · Amazon (United States)
- Evaluates
- Agents, Cybersecurity, Language & Knowledge, Robotics & Embodied AI
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
SEVRA-BENCH paper
No reported scores are on record for this benchmark yet.