Claude Skill
MrGeDiao/shuorenhua
Shuorenhua is a Chinese-first AI writing refinement skill that reduces boilerplate and AI tone while preserving facts, terminology, and technical context. Open-source and compatible with Claude, Ch...
Overview
Repository
Install this Skill
git clone https://github.com/MrGeDiao/shuorenhua.git && cd shuorenhuaRegistry
git clone https://github.com/MrGeDiao/shuorenhua.git && cd shuorenhuagit clone https://github.com/MrGeDiao/shuorenhua.git
Summary
Shuorenhua (说人话) is a Chinese-first AI writing refinement skill that reduces boilerplate and AI tone while preserving facts, terminology, and technical context. It helps users produce more natural, human-like Chinese text from AI outputs.
说人话|中文优先的AI写作优化技巧,减少套话和AI腔调,同时保留事实、术语和技术语境。
Key features
- Chinese-first refinement for AI writing
- Reduces boilerplate and AI tone
- Preserves facts, terminology, and technical context
- Compatible with Claude, ChatGPT, Cursor, and more
- Open-source prompt engineering skill
Use cases
- Polishing AI-generated Chinese articles
- Improving naturalness in technical documentation
- Refining chatbot or assistant responses in Chinese
- Reducing robotic tone in professional writing
- Enhancing readability of AI outputs for Chinese readers
README excerpt
<h1 align="center">说人话:中文 AI 味清理 skill</h1> <p align="center"> <picture> <source media="(prefers-color-scheme: dark)" srcset="assets/banner-dark.svg"> <img src="assets/banner-light.svg" alt="说人话:中文 AI 味清理 skill — 先保信息,再谈风格" width="100%"> </picture> </p> <p align="center"> <strong>别让模型替你装腔。</strong> </p> <p align="center"> 给 Codex、Claude Code、Cursor、ChatGPT 和自建 agent 用。 <br> 改聊天、技术同步、README、论坛帖和中文长文:先保住事实,再把那股“一眼 AI”的腔调降下来。 </p> <p align="center"> <a href="https://github.com/MrGeDiao/shuorenhua/stargazers"><img src="https://img.shields.io/github/stars/MrGeDiao/shuorenhua?style=for-the-badge&label=stars" alt="GitHub stars"></a> <a href="https://github.com/MrGeDiao/shuorenhua/releases"><img src="https://img.shields.io/github/v/release/MrGeDiao/shuorenhua?style=for-the-badge&label=release" alt="GitHub release"></a> <a href="evals/benchmark.md"><img src="https://img.shields.io/badge/benchmark-72%20cases-2563eb?style=for-the-badge" alt="Benchmark: 72 cases"></a> <a href="evals/real-samples.md"><img src="https://img.shields.io/badge/real%20samples-19-16a34a?style=for-the-badge" alt="Real samples: 19"></a> <a href="LICENSE"><img src="https://img.shields.io/github/license/MrGeDiao/shuorenhua?style=for-the-badge" alt="License"></a> </p> <p align="center"> <a href="#改成什么样">改成什么样</a> · <a href="#30-秒上手">30 秒上手</a> · <a href="#它怎么判断怎么改">怎么改</a> · <a href="#评测">评测</a> · <a href="#安装">安装</a> · <a href="#常见问题">FAQ</a> </p> `说人话` 专治那种“每个字都对,但一看就不是你写的”中文。它不把空话包装得更漂亮,也不替你编新事实;它先护住版本、命令、责任和证据,再拆掉过度承接、工程师腔、小红书 AI 腔、翻译腔和无源权威铺垫。目标很简单:改完你敢直接发。 它适合这些场景: | 场景 | 它会做什么 | |------|------------| | 日常聊天 | 删掉过度承接、推销式结尾和工程汇报腔,保留口语感 | | 技术状态同步 | 保住事实、版本、命令、报错和责任归属,压低套话 | | README / release note | 先讲清楚项目、变更、验证和限制,不写发布宣言 | | 论坛帖 / issue 回复 | 像维护者在认真