List your company

Malicious-skill detection on 906-skill benchmark

94.6F1 ScoreBehavioral integrity (BIV)
41.97655.63869.382.962Jun 30, 2026

Evaluation results

MethodLinks
Behavioral integrity (BIV)axis=behavior2026.0694.697.87.2
LLM-only auditaxis=behavior2026.0693——
Structural fingerprintaxis=identity2026.0673.58132
Rule-based scanneraxis=n/a2026.0644——