hunt-clickjacking

This skill is internally coherent and not overtly malicious, but it gives an AI agent offensive web-testing capability against real targets and sensitive actions. Main risk comes from exploit-oriented use, not supply chain, credential theft, or covert exfiltration.