ai-detect-skill

Audit drafts, slides, reports, homework, and sendable external messages for AI-smelling wording using Mingdao's historical session feedback. Use when the user asks whether a deliverable has 'AI味', sounds unnatural, has high entropy, uses placeholder titles, over-explains, contains process filler, or does not sound like a normal human wrote it.

Audit bilingual formal deliverables for low-information repetition, empty contrast, sequencer scaffolds, adjacent restatement, and repeated comparison-axis labels. Use when the user asks whether a draft feels redundant, wordy, low-information, or says the same thing twice.