security-auditor-agent

This is a coherent security-audit skill, but it grants an AI agent offensive security assessment capability, including active probing and exploit-style testing. That makes it high risk as an agent skill despite no clear malware, credential theft, or suspicious exfiltration paths.