audit-support
Warn
Audited by Socket on Mar 23, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: The core audit-analysis behavior is mostly consistent with the stated purpose and has no evident external exfiltration or credential harvesting. Risk comes from autonomous broad local scanning, writing to a home-directory memory location, and especially the references to additional unverified slash commands, which create a transitive trust chain disproportionate to a single audit-support skill.
Confidence: 83%Severity: 56%
Audit Metadata