variant-analysis

SUSPICIOUS: the skill is internally coherent and does not show credential theft or malicious data flows, but it grants an AI agent explicit vulnerability-hunting capability across codebases. That offensive-security purpose makes it high risk despite otherwise clean install and data-flow characteristics.