pentest-exploit-validation

SUSPICIOUS: internally coherent as an offensive pentest skill, but it gives an AI agent high-risk exploitation capabilities, including bypass automation, data exfiltration, account takeover validation, and SSRF metadata access. The main concern is not deceptive purpose mismatch; it is that the skill operationalizes real attack techniques with meaningful real-world impact and forwards sensitive credentials/tokens into several external tools.