j-idol

This document is a high-risk instruction template for generating photorealistic, highly sexualized gravure portraits derived from a reference image while preserving identity-linked attributes. The greatest threats are non-consensual sexual-image generation (deepfake risk), potential sexualization of minors (no age checks), and privacy violations from sending identifiable reference images to external services. There is no evidence of embedded malware or obfuscation in the fragment, but absence of consent, age verification, refusal rules, and secure handling of reference images makes this skill unsafe to run as-is. Recommended mitigations: block or require strict enforcement before use — demand explicit, auditable consent from the subject; implement age-verification; refuse requests involving public figures or private individuals without proof of consent; avoid embedding exact identity attributes in prompts; add automatic watermarking and retention/secure-storage policies; audit outbound API calls and logging to prevent exfiltration.