p-video-avatar
Warn
Audited by Socket on Jul 23, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The avatar-generation purpose is coherent, and sending media inputs to inference.sh is proportionate, but the trust chain is weaker than it should be: the skill asks to install an external CLI via an unverified `belt-sh/cli` target, links to mutable raw GitHub install content, grants shell access for belt commands, and encourages transitive skill installation. This looks more like supply-chain and trust-expansion risk than confirmed malicious behavior.
Confidence: 85%Severity: 72%
Audit Metadata