Clean Chinese AI-sounding writing with Shuorenhua
Rewrite Chinese-first chat, status, docs, and public-writing drafts to remove AI-like phrasing while preserving facts, commands, numbers, terms, and responsibility.
npx skills add agentskillexchange/skills --skill clean-chinese-ai-sounding-writing-with-shuorenhua
/plugin marketplace add MrGeDiao/shuorenhua and /plugin install shuorenhua@shuorenhua. For Codex, clone the repo and run codex exec -C . "读取 ./SKILL.md，按其中规则改写以下文本：...". Runtimes with the skills CLI can run npx skills add MrGeDiao/shuorenhua.Use Shuorenhua when an agent is asked to make Chinese or mixed Chinese-English text sound more natural, less template-like, or less visibly AI-written while preserving the original facts. The operator loads the packaged `SKILL.md`, detects the text scene, protects numbers, commands, terms, quotes, and responsibility, then rewrites or annotates the draft under documented scope rules. Invoke this instead of using a normal writing tool when the task is style cleanup with fact-preservation constraints, especially for Chinese chat replies, status updates, READMEs, release notes, forum posts, or issue replies. The boundary is the reusable anti-AI-tone editing workflow with references and evals; it is not a generic writing app, prompt collection, or broad translation tool.
What this skill actually does
Inputs and prerequisites: Codex, Claude Code, Cursor, ChatGPT, OpenClaw, or another agent runtime that can load a SKILL.md file.
Setup notes: For Claude Code, run /plugin marketplace add MrGeDiao/shuorenhua and /plugin install shuorenhua@shuorenhua . For Codex, clone the repo and run codex exec -C . “读取 ./SKILL.md，按其中规则改写以下文本：…” . Runtimes with the skills CLI can run npx skills add MrGeDiao/shuorenhua .
Source and verification boundary: use https://github.com/MrGeDiao/shuorenhua as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.