ai-redteam

SUSPICIOUS. The skill’s capabilities are broadly aligned with its stated purpose as an AI red-teaming guide, and the supplied text shows no installers, credential requests, remote endpoints, or obvious exfiltration paths. However, it explicitly equips an AI agent for offensive-style security testing of AI systems, which is high-risk by nature, and the unseen reference files could materially change behavior. Based on the provided content alone, this looks like a documentation/workflow skill with elevated misuse potential rather than confirmed malware.