judge-pentest
Warn
Audited by Socket on Apr 22, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS: the skill is internally consistent as a pentest evaluator, but its actual purpose is to help an AI agent direct offensive security activity. No clear credential theft or exfiltration is shown, yet AI-agent exploit guidance and continued testing logic make it high security risk.
Confidence: 92%Severity: 83%
Audit Metadata