google-veo
Warn
Audited by Socket on May 13, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the core function is coherent, but the skill depends on a same-org remote-installed CLI, sends Veo requests through inference.sh instead of direct Google APIs, uses relatively broad `belt *` permissions, and encourages transitive skill installation. This looks more like a hosted platform wrapper than a direct Google Veo integration; risky but not confirmed malicious.
Confidence: 87%Severity: 56%
Audit Metadata