humanize-with-rephrasy
Warn
Audited by Socket on Oct 20, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The install source appears same-org and the network endpoints match vendor docs, so this is not obvious malware or a fake gateway. However, the skill's core purpose is detector evasion, it recommends unpinned remote package execution, and it forwards an API key plus user text through third-party code and a remote service. Internally coherent, but high-risk and inappropriate for many environments.
Confidence: 88%Severity: 68%
Audit Metadata