ai-red-teaming

SUSPICIOUS. The skill is coherent for its stated purpose, but that purpose is offensive security testing for AI agents, which carries elevated misuse risk by design. There is no clear credential harvesting, covert exfiltration, hidden execution, or untrusted installer in the skill text itself; the main concern is that it equips an agent with attack-planning guidance, even though it includes strong authorization, sandboxing, and no-real-side-effects guardrails.