review
Warn
Audited by Socket on Nov 8, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the core PR review behavior is legitimate, and the helper binaries appear same-publisher rather than an obvious third-party credential trap. But the skill's footprint is much broader than 'pre-landing review': it executes many local binaries, uses eval/source on their outputs, can edit and commit workflow files, runs external AI/tooling, and performs optional telemetry/artifact sync. This is more of a full gstack orchestration skill than a narrowly scoped review skill, so the capability scope is only partially aligned and carries medium risk.
Confidence: 88%Severity: 62%
Audit Metadata