Chinese De-AI Styling And Literary Enhancement

Problem

Chinese AI-generated text often reads smooth but flat: too orderly, full of abstract nouns, false balance, and generic endings. Readers may finish without remembering any judgment. Worse, many drafts carry content debt, not only language debt: missing facts, real scenes, reader tasks, or supported opinions. Simple polishing can hide these gaps. This skill is useful for editing Chinese social posts, Xiaohongshu notes, WeChat articles, essays, reports, proposals, and product copy, pulling drafts back to concrete people, decisions, and evidence.

How It Works And Limits

It separates diagnosis from rewriting. For diagnosis only, it lists high-impact issues in order, quotes short fragments, and explains why each one weakens the piece. For rewriting, it first identifies content type and the reader's task, then distinguishes content debt from language debt. For content debt, it asks for facts, examples, trade-offs, data, or actions. For language debt, it removes stock AI patterns, inflated adjectives, consulting templates, and overused connectors, and adds clarity through naming, visible imagery, rhythm, and action. It can also read files such as references/anti-ai-rules.md and references/literary-lift.md when the task needs diagnosis or literary lift. Limits: it does not roleplay a persona, fabricate sources, add slang or exaggerated tone, or invent credentials, numbers, and cases. When the original draft lacks a core judgment, it works better as an editorial check than as a generator of new opinions.