hunt-llm-ai

This skill is a high-risk offensive security playbook for AI agents, not a narrowly scoped defensive reference. Its actual footprint centers on exploit development, system prompt extraction, credential/data exfiltration, use of known OAST endpoints, and live-target compromise across LLM apps, MCP servers, vector DBs, and model servers. No strong malware proof is embedded, but the skill is clearly dangerous and should be classified as maliciously capable/high-risk.