draft-pr
Warn
Audited by Socket on Sep 15, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
The core GitHub/gh workflow is coherent and mostly proportionate for a PR-publishing skill, with direct use of official GitHub tooling and no obvious credential exfiltration. The main concerns are transitive trust in a remotely fetched sibling skill from a personal GitHub repo and the directive to hide agent attribution in published artifacts; together these make the skill suspicious rather than benign, but not confirmed malware.
Confidence: 86%Severity: 56%
Audit Metadata