babysitting-pr
Warn
Audited by Socket on Sep 14, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the skill’s capabilities broadly match its PR-maintenance purpose, and it uses official GitHub tooling rather than third-party credential relays. The main risk is workflow autonomy: it reads untrusted PR comments/CI logs and can modify code, push commits, and change PR state with limited approval, creating a moderate indirect prompt-injection and real-world action surface.
Confidence: 89%Severity: 61%
Audit Metadata