DDIPE embeds malicious logic within the legitimate technical structures of skill documentation — not as standalone behavioral instructions. The two embedding strategies exploit how coding agents treat documentation as authoritative reference for code synthesis.