List your company

Malicious Skill Detection on MaliciousAgentSkillsBench (404 malicious, 502 benign)

100RecallBIV
32.449.9567.585.05May 12, 2026

Evaluation results

MethodLinks
BIVModel=Claude Sonnet 4.52026.051009999
BIVModel=Claude Sonnet 4.52026.05989295
LLM-only2026.05939996
LLM-only2026.05899793
BIVModel=Claude Sonnet 4.52026.05865769
LLM-only2026.05647468
Rule-based2026.05458860
Rule-based2026.05392026
Rule-based2026.05356244