skills-store
Warn
Audited by Socket on Apr 23, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The skill’s capabilities match its purpose and its network/data flow stays within PostHog, so it is not malware-like. The main risk is architectural: it instructs the agent to load remote shared skill bodies/files and treat them as instructions, creating meaningful indirect prompt-injection and transitive-trust risk.
Confidence: 89%Severity: 64%
Audit Metadata