video-model-selection
Warn
Audited by Socket on Sep 8, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The core skill content is coherent and APImage routing is same-service and officially documented, but the Codex setup introduces a meaningful supply-chain and credential-forwarding risk by using unpinned third-party `mcp-remote` with the APImage API key. This is not confirmed malware, but it exceeds the minimal trust footprint needed for a model-selection guidance skill.
Confidence: 90%Severity: 62%
Audit Metadata