pre-ship-review
Warn
Audited by Socket on Apr 4, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: The skill’s core review behavior is coherent and mostly benign for a pre-ship QA workflow, but its footprint is broadened by transitive skill orchestration and self-modification instructions. The main concrete security issue is install trust: the documented `pip install pyright` path uses an unofficial third-party wrapper rather than the primary publisher distribution, making the toolchain less trustworthy than the skill implies. No credential harvesting, exfiltration endpoints, or clearly malicious data flows are present.
Confidence: 87%Severity: 56%
Audit Metadata