external-model
Warn
Audited by Socket on Jun 23, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS from a security-risk perspective, not because the purpose is deceptive, but because the skill brokers prompts and optional file context into external authenticated AI CLIs and can grant them repo write access. Purpose and capabilities are mostly aligned, and the known installers are official same-org paths, so this is not confirmed malware; however, it materially expands trust to third-party binaries and model providers.
Confidence: 83%Severity: 56%
Audit Metadata