varg-ai
Warn
Audited by Socket on Apr 27, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The core media-generation behavior and official varg.ai data flows are coherent with the stated purpose, so this does not look like outright malware. Risk comes from direct credential-file handling, writing secrets into .env, remote version checking plus transitive npx skill updates, and the ability to initiate billing checkout flows using an OTP-derived access token.
Confidence: 83%Severity: 58%
Audit Metadata