gitmyhub

stop-slop-zh

★ 28 updated 17d ago

🚫 Stop AI slop in Chinese writing. 干掉中文 AI 味 — banned words, punctuation rules, structure constraints & 4-layer QA. Works with Claude Code, Cursor, Codex CLI.

A writing rule set you load into AI coding tools to stop them from producing robotic-sounding Chinese text, replacing filler phrases and stiff structure with natural, conversational language.

setup: easycomplexity 1/5

Stop Slop ZH is a writing rule set designed to remove the patterns that make AI-generated Chinese text feel machine-written. It is not a standalone application, but a skill file that you load into AI tools like Claude Code, Cursor, or Codex CLI to change how they generate Chinese prose.

Chinese AI writing tends to fall into recognizable patterns: connecting phrases like "it is worth noting that" or "in summary," numbered list structures, heavy use of colons and double quotes, and a formal newsreader tone. This project identifies those patterns and instructs the AI to avoid them, replacing them with more conversational and specific language that sounds like a person writing from direct experience.

The project consists of a main instruction file (SKILL.md) and a set of reference files. The reference files catalog over 30 banned filler phrases, banned punctuation marks, banned structural patterns (such as first/second/third enumerations), and a library of over 50 recommended colloquial Chinese expressions to use instead. There is also a before/after example file showing the difference between typical AI output and the revised version.

When the AI finishes writing or revising text, it runs a four-layer quality check: verifying hard punctuation and word rules, reviewing style, checking whether the content has enough concrete detail, and assessing whether the text reads like a person rather than a machine. It produces a short report after each check.

Installation is a single git clone into your AI tool's skills directory. The project is MIT licensed and was built by a content team that uses the same rules in their own daily Chinese writing work.

Where it fits