performing-penetration-testing

SUSPICIOUS: the skill is purpose-consistent and not obviously malicious, but it equips an AI agent to conduct penetration-testing workflows with broad execution scope and transitive trust in 25 subordinate skills. The main concerns are offensive-security capability, autonomous orchestration, and partial inconsistency in enforcing authorization-first behavior; supply-chain indicators in this file are relatively low.