attack-tree-construction
Warn
Audited by Socket on Apr 14, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS due to offensive-security enablement for an AI agent, even though the skill is otherwise tightly scoped and instruction-only. No credential access, installs, or external data flows are present; risk comes mainly from giving the agent structured attack-modeling capability.
Confidence: 90%Severity: 58%
Audit Metadata