Chinese Writing De-Slop Guide

Problem

Chinese technical docs, business writing, and marketing copy often carry machine-like habits: padded openings, stacked connectors, evenly balanced sentences, and template-like metaphors. Each sentence may be grammatical, but the whole reads like an AI draft. The editing problem is not syntax; it is recognizing over-polished phrasing, weak stance, and surface-level fluency.

How It Works

The skill turns de-slop editing into a repeatable process: scan the text, rewrite against explicit rules, then audit for newly introduced clichés. Core capabilities include:
- Detecting 58 Chinese AI-trace patterns across filler phrases, structural formulas, lexical alerts, formatting abuse, narrative tone, business-document habits, chat residue, and register issues.
- Applying five rewrite rules: removing low-information phrases, breaking formulaic structures, varying sentence rhythm, stating facts directly, and replacing slogan-like lines.
- Adding a human voice where useful: taking a position, acknowledging complexity, using first person sparingly, and making feelings concrete.
- Using a quick checklist and a five-dimensional score to decide whether the result is acceptable, with another pass below 35/50.

The output is a rewritten Chinese text plus brief change notes. It is useful for polishing LLM drafts, long-form posts, feasibility reports, and technical blogs.

Boundaries

This skill focuses on Chinese stylistic editing, not fact-checking, data completion, or typesetting. Thresholds for formal memos, book manuscripts, news, and academic writing differ; the skill evaluates whether bold text, quotation marks, lists, and colloquial words are excessive for the register. A single dash, one triadic list, or a formal tone should not be treated as proof of AI writing. The final rewrite should still preserve the author's position and factual boundaries.