model-extraction

SUSPICIOUS: the skill is internally consistent, but its purpose is to equip an AI agent with offensive model-theft capabilities and evasion tactics. There is little supply-chain or credential risk, yet the operational impact is high because it automates exploit-style extraction against external AI services.

This YAML configuration describes model extraction and includes explicit measures to evade query monitoring and rate limits. It contains no executable behavior or direct malware indicators, but presents a significant misuse risk if applied to a model without authorization. No domains, URLs, IP addresses, file paths, or credentials are present.