claude-review
Warn
Audited by Socket on Apr 21, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the core purpose is coherent and the Claude CLI comes from an official same-org source, but the skill grants Claude `bypassPermissions` and transitively invokes unverified third-party skills in the same session. Risk is driven more by excessive execution trust and inherited permissions than by confirmed malicious behavior.
Confidence: 86%Severity: 68%
Audit Metadata