Agent skill

Huashu Proofreading

by alchaincyf in alchaincyf/huashu-skills

三遍审校降低AI检测率,让文章更有人味。当用户提到"AI味太重"、"像AI写的"、"降低AI检测率"、"审校"、"自然一些"时使用。

MITAuto-check passedWriting & Content

Install Huashu Proofreading

skills CLI
$ npx skills add alchaincyf/huashu-skills --skill huashu-proofreading -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install alchaincyf/huashu-skills huashu-proofreading --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/alchaincyf/huashu-skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/huashu-proofreading .claude/skills/huashu-proofreading && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
huashu-proofreading
GitHub stars
1.7k
Token cost
~1.2k tokens
SKILL.md length
209 words
Files
1
Skills in repo
21
Repo updated
First seen
Licence
MIT

At a glance

三遍审校降低AI检测率,让文章更有人味。当用户提到"AI味太重"、"像AI写的"、"降低AI检测率"、"审校"、"自然一些"时使用。

  • Works in 5 steps: 逐段检查,标注AI味表达 → 对照6大类AI腔清单改写 → 加入个人经历和真实细节(查个人素材库) → …
  • Tasks that involve Copy editing and proofreading
  • SKILL.md covers 何时使用此Skill, 核心目标, 三遍审校流程 and 花生风格特征(必须保持), plus 5 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Huashu Proofreading is an agent skill from alchaincyf/huashu-skills. 三遍审校降低AI检测率,让文章更有人味。当用户提到"AI味太重"、"像AI写的"、"降低AI检测率"、"审校"、"自然一些"时使用。

Its SKILL.md is about 1.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Writing & Content, covering Copy editing and proofreading. The repository describes itself as: 花叔全部开源 Agent Skills 总目录:16 旗舰 + 14 人物视角 + 22 内置共 52 个 skill,分层分类 + AI Agent 安装协议 + 机器可读 skills.json + 更新检查机制. The licence is MIT.

When your agent uses it

  • Tasks that involve Copy editing and proofreading

Example prompts

  • “降低AI检测率”
  • “/huashu-proofreading”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. 逐段检查,标注AI味表达
  2. 对照6大类AI腔清单改写
  3. 加入个人经历和真实细节(查个人素材库)
  4. 加入不确定性表达(适度使用"我觉得"、"可能"、"也许")
  5. 检查是否像花生在说话

What it can do on your machine

Read from SKILL.md and the folder at commit 2efea35. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Huashu Proofreading loads about 1.2k tokens when it runs. Until then it costs about 22 tokens; SKILL.md has 209 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~22
When it runs · the whole SKILL.md, loaded when a task matches
~1.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from alchaincyf/huashu-skills at commit 2efea35, republished under its MIT licence (© alchaincyf). 209 words, ~1,226 tokens.

Download SKILL.mdSave it as .claude/skills/huashu-proofreading/SKILL.md (or your agent's skills folder).
name
huashu-proofreading
description
三遍审校降低AI检测率,让文章更有人味。当用户提到"AI味太重"、"像AI写的"、"降低AI检测率"、"审校"、"自然一些"时使用。

AI味审校

系统化降低文章AI检测率,增加人味的三遍审校专业能力。

何时使用此Skill

当检测到以下场景时,自动加载此skill:

  • 用户说"审校一下"、"降低AI味"、"太AI了"
  • 初稿创作完成后
  • 用户提到"AI检测率高"、"没人味"
  • 文章需要最终polish

核心目标

  • 降低AI检测率:目标控制在30%以下
  • 增加人味:让文章有经验感、有温度、像真人在说话
  • 保持花生风格:实践导向、技术深度、易懂表达、有用可落地

三遍审校流程

第一遍:内容审校(逻辑、事实、结构)

目标:确保内容准确、逻辑清晰

检查项:

  • 事实准确?(数据、时间、产品名称)
  • 逻辑清晰?(前后无矛盾)
  • 结构合理?(无跑题)
  • 信息完整?(关键点都说了)
  • 无编造?(所有数据和案例都是真实的)

方法:

  • 从头到尾读一遍,标注疑问点
  • 搜索验证不确定的信息
  • 检查数字和专有名词的准确性

第二遍:风格审校(AI味降重)

目标:去掉AI味,增加人味

核心检查:6大类AI腔识别

1️⃣ 套话连篇

❌ AI常说:

  • "在当今时代"、"在...的大背景下"
  • "综上所述"、"总而言之"
  • "值得注意的是"、"需要强调的是"
  • "随着...的发展"、"伴随着..."

✅ 改写方向:直接切主题,不绕弯子

示例:

❌ "在当今AI技术飞速发展的时代,编程工具也在不断进化..."
✅ "Claude Code出了。我用了两周,确实比Cursor好用。"
2️⃣ AI句式(高频重复模板)

❌ AI常说:

  • "不是...而是..."(连续出现3次以上)
  • "不仅...而且..."(堆叠使用)
  • "一方面...另一方面..."
  • "既...又..."(过度使用)

✅ 改写方向:拆成短句,用多样化表达

示例:

❌ "Claude Code不是简单的工具,而是智能助手。它不仅能写代码,而且能理解意图。"
✅ "Claude Code不只是补全代码。它能理解你的意图,知道你想做什么。这个很关键。"
3️⃣ 书面词汇(距离感强)

❌ AI常说:

  • "显著提升"、"大幅改善"→ 用具体数字
  • "充分利用"→ "用好"
  • "进行操作"→ 直接用动词("点击"、"输入")
  • "实现功能"→ "做到"
  • "促进发展"→ "帮助..."

✅ 改写方向:口语化,具体化

示例:

❌ "通过充分利用Claude Code的能力,可以显著提升开发效率..."
✅ "用好Claude Code,你的开发速度能快不少。我测下来,平均每个项目省3-5小时。"
4️⃣ 结构机械(AI式列表癖)

❌ AI特征:

  • 过度使用"首先、其次、最后"
  • 无意义的小标题密集(每200字一个)
  • 能用散文讲清楚的,非要列表
  • "以下几点"、"主要体现在以下方面"

✅ 改写方向:自然叙事,列表适度

示例:

❌ "Claude Code的优势主要体现在以下几个方面:首先,它具有强大的上下文理解能力;其次,它支持多种编程语言;最后,它的响应速度很快。"

✅ "Claude Code最强的地方是什么?理解上下文。你让它改一个功能,它知道要动哪几个文件,不会漏掉。而且支持20多种语言,速度也很快。"
5️⃣ 态度中立(无立场无观点)

❌ AI特征:

  • 永远客观平衡:"既有优点也有缺点"
  • 不敢下判断:"具体取决于实际情况"
  • 每句话都加限定词:"可能"、"或许"、"在某些情况下"

✅ 改写方向:明确态度,敢下判断

示例:

❌ "Claude Code在某些场景下可能表现较好,但也存在一些局限性,具体效果取决于实际使用情况。"

✅ "Claude Code确实好用。处理大型项目时,比Cursor稳定多了。但价格贵,个人开发者可能扛不住。"
6️⃣ 细节缺失(抽象不落地)

❌ AI特征:

  • 大词多,具体细节少
  • "许多"、"一些"、"显著"、"大量"(不给数字)
  • 缺少真实感受和个人经历
  • 无时间线、无场景、无冲突

✅ 改写方向:加入真实细节、具体数字、个人经历

示例:

❌ "经过一段时间的使用,我发现Claude Code能够有效提升开发效率。"

✅ "用了两周。最明显的变化是,之前要3天的项目,现在1天半就能搞定。特别是重构老代码,Claude Code能一次性找出所有关联文件,不用我一个个翻。"

第二遍审校:具体方法
  1. 逐段检查,标注AI味表达
  2. 对照6大类AI腔清单改写
  3. 加入个人经历和真实细节(查个人素材库)
  4. 加入不确定性表达(适度使用"我觉得"、"可能"、"也许")
  5. 检查是否像花生在说话

常用改写技巧:

  • 长句拆短句(超过30字的拆成2-3句)
  • 抽象改具体("很多" → "3个","显著提升" → "省3-5小时")
  • 书面改口语("进行操作" → "点击")
  • 加入真实感受("我之前用的时候..."、"说实话...")

第三遍:细节打磨(标点、排版、节奏)

目标:让文章读起来舒服、自然

检查项:

  • 句子长度合适?(15-25字为主,不超过30字)
  • 段落不太长?(手机屏幕3-5行)
  • 标点自然?(多用句号,少用逗号连接长句)
  • 节奏有变化?(快慢结合)
  • 小标题不密集?(每300-500字一个)
  • 列表适度?(能用散文就不用列表)
  • 加粗适度?(每200-300字1-2处)

方法:

  • 大声朗读,感受节奏
  • 找出超过30字的长句,拆短
  • 检查段落长度,过长的分段
  • 调整排版,删除过度的格式化

最终检查
  • 从头到尾读一遍,卡顿的地方改写
  • 感觉像花生本人在说话
  • (可选)AI检测率低于30%

花生风格特征(必须保持)

  • 实践导向:有具体案例、可执行步骤
  • 技术深度 + 易懂表达:不掉书袋,但有深度
  • 有用可落地:读者看完能立刻行动
  • 人味浓:有经验感、有温度,避免AI腔
  • 数据真实:所有测试、案例都必须真实
  • 适度不确定性:该说"我觉得"就说,不假装全知

个人素材库使用

何时使用:

  • ✅ 文章需要真实案例支撑
  • ✅ 需要对工具/产品的真实评价
  • ✅ 需要个人经历作为开场白
  • ✅ 文章太"AI味",需要增加人味

快速使用流程:

方法A:直接搜索原始数据(推荐优先使用)

适用场景:新话题、未整理的主题、需要最原始的真实表达

使用Grep工具搜索 /写作参考/个人素材库/全部即刻动态.csv:

pattern: "关键词1|关键词2"
path: /Users/alchain/Documents/写作/写作参考/个人素材库/全部即刻动态.csv
output_mode: content
-n: true

示例:

  • 写高德扫街榜 → 搜索"高德|扫街榜"
  • 写AI编程工具 → 搜索"Cursor|Claude Code|Copilot"
  • 写产品增长 → 搜索"增长|转化|用户"
方法B:查看已提炼素材

适用场景:常见主题(AI编程工具、产品开发等)已有系统化整理

  1. 打开总索引:/写作参考/个人素材库/README.md
  2. 查找主题:打开 _即刻动态主题索引.md
  3. 选择素材:查看提炼文件(如AI编程工具使用心得.md)
  4. 改写使用:扩展成长文逻辑,保持真实态度

注意事项:

  • 所有素材都是真实的,不能编造或夸大
  • 注意时间跨度(2025.05-2025.10),避免过时信息
  • 必须改写成长文逻辑,不能直接复制粘贴

审校工具(可选)

自动检测工具
  • 关键词检测:搜索"在当今"、"综上所述"、"值得注意的是"等套话
  • 句长检查:标注超过30字的长句
  • AI检测:使用GPTZero、Originality.AI等工具(仅供参考)
手动检测方法
  • 大声朗读:卡顿的地方就是需要改的地方
  • 对比历史文章:和花生之前的优秀文章对比,看风格是否一致
  • 问自己:这句话花生会这么说吗?

常见错误与改写示例

错误1:开头绕圈子
❌ "在当今AI技术飞速发展的背景下,编程工具领域也迎来了新的变革。作为一款新兴的AI编程助手,Claude Code正在受到越来越多开发者的关注..."

✅ "Claude Code出了。我用了两周,确实比Cursor好用。今天聊聊为什么。"
错误2:过度使用AI句式
❌ "Claude Code不是简单的代码补全工具,而是一个智能编程助手。它不仅能够生成代码,而且能够理解开发者的意图。"

✅ "Claude Code不只是补全代码。它能理解你想干嘛。比如你说'重构这个模块',它就知道该怎么做。"
错误3:缺少具体细节
❌ "经过一段时间的测试,我发现Claude Code在很多方面都表现优秀。"

✅ "用了两周。最明显的变化是,之前要3天的项目,现在1天半就能搞定。"
错误4:态度过于中立
❌ "Claude Code有其优势,也存在一些不足,具体是否适合使用需要根据实际情况判断。"

✅ "Claude Code确实好用。但价格贵,个人开发者可能扛不住。如果你公司报销,强烈推荐。"

参考资源

  • /写作参考/AI味审校指南.md - 完整的审校指南(包含更多示例)
  • /写作参考/个人素材库/README.md - 个人素材库总索引
  • /写作参考/个人素材库/全部即刻动态.csv - 原始素材数据
  • /写作参考/个人素材库/_即刻动态使用规范.md - 素材使用规范

成功案例

  • 《为什么最好的内容都是无用的》:经过三遍审校,AI检测率降至28%
  • 《一个月烧几十亿tokens的团队,是怎么用Claude Code的》:加入大量真实案例,人味浓厚

最后更新: 2025-11-07 适用项目: 公众号写作、视频脚本 维护者: 花生


花叔出品 | AI Native Coder · 独立开发者 公众号「花叔」| 30万+粉丝 | AI工具与效率提升 代表作:小猫补光灯(AppStore付费榜Top1)·《一本书玩转DeepSeek》

© alchaincyf, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in huashu-proofreading of alchaincyf/huashu-skills.

Open the folder on GitHubat commit 2efea35

Compare with similar skills

Huashu Proofreading next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Huashu Proofreading compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Huashu Proofreading this skillalchaincyf/huashu-skills1.7k—~1.2kAutomated safety check: PassMIT
User-Facing Text Cleanupguillaumemeyer/watermarks-remover24k—~3.5kAutomated safety check: PassMIT
Story Multi-Perspective Reviewzenstory-ai/oh-story-claudecode7.4k3 repos~3kAutomated safety check: PassMIT
Chinese Text Humanizerop7418/Humanizer-zh19k—~2kAutomated safety check: PassMIT
Baoyu TranslateJimLiu/baoyu-skills27k1 repos~3.9kAutomated safety check: PassMIT
Natural Japanese Business Writingcoji/natural-japanese1.9k—~2.1kAutomated safety check: PassMIT

Similar skills

  • User-Facing Text Cleanup

    guillaumemeyer/watermarks-remover

    Audits prose for invisible Unicode characters and rewrites it while keeping facts, citations, code and required disclosures unchanged and the writer's voice intact.

    24k GitHub stars~3.5k tokensUpdated yesterday
    Writing & ContentAuto-check passed
  • Story Multi-Perspective Review

    zenstory-ai/oh-story-claudecode

    Reviews Chinese web-novel text with several reviewer agents in parallel, falling back to a single-agent pass, and reports structure, character, prose and setting problems with fixes.

    7.4k GitHub starsUsed in 3 repos~3k tokens
    Writing & ContentAuto-check passed
  • Chinese Text Humanizer

    op7418/Humanizer-zh

    Edits Chinese articles, comments and documents to remove filler, repetition and template phrasing while keeping the facts, the level of certainty and the author's voice.

    19k GitHub stars~2k tokensUpdated 17 days ago
    Writing & ContentAuto-check passed
  • Baoyu Translate

    JimLiu/baoyu-skills

    Translates articles and files in quick, normal or refined mode, with a custom glossary, saved preferences and a review-and-polish workflow for publication quality.

    27k GitHub starsUsed in 1 repo~3.9k tokens
    Writing & ContentAuto-check passed
  • Writes and edits Japanese business documents so they read clearly and naturally, removes AI-sounding phrasing and can score how AI-like a text reads.

    1.9k GitHub stars~2.1k tokensUpdated 1 mo ago
    Writing & ContentAuto-check passed
  • Defensive Writing Editor

    lennney/stop-that-shit

    Cuts defensive disclaimers, stacked hedging and self-protective narration from proposals and summaries, keeping only limits that affect the reader's decision.

    2.5k GitHub starsUsed in 1 repo~1.2k tokens
    Writing & ContentAuto-check passed

More from alchaincyf/huashu-skills

All 21 skills in this repo
  • Huashu Agent Swarm

    alchaincyf/huashu-skills

    多Agent蜂群并行协作,纯git自组织,适合大型项目开发。当用户提到"蜂群模式"、"多agent"、"并行开发"、"agent swarm"时使用。

    1.7k GitHub stars~576 tokensUpdated 17 days ago
    Auto-check passed
  • Huashu Data Pro

    alchaincyf/huashu-skills

    数据分析与办公提效全能助手。覆盖数据处理、分析洞察、报告撰写、PPT制作、数据可视化的端到端工作流. An agent skill from alchaincyf/huashu-skills.

    1.7k GitHub stars~928 tokensUpdated 17 days ago
    Auto-check passed
  • Huashu Douyin Script

    alchaincyf/huashu-skills

    抖音爆款脚本创作工作流。从竞品视频拆解到脚本生成的完整流程:下载抖音视频→Gemini视频分析→爆款公式提炼→脚本+分镜生成→AI味审校。

    1.7k GitHub stars~1.7k tokensUpdated 17 days ago
    Auto-check passed
  • Huashu Wechat Image

    alchaincyf/huashu-skills

    为微信公众号文章生成高质量配图。支持封面图(2.35:1)、正文插图(16:9/4:3)、信息图。提供两条路径:AI生成(视觉创意型)和HTML渲染(文字精确型)。当用户提到"公众号配图"、"公众号封面"、"文章配图"、"正文插图"、"公众号图片"时使用此技能。

    1.7k GitHub stars~2.4k tokensUpdated 17 days ago
    Auto-check: notes
  • Huashu Xhs Image

    alchaincyf/huashu-skills

    为小红书笔记生成高质量配图。默认AI生成(Gemini),仅精确数据表格用HTML兜底。当用户提到"小红书配图"、"小红书封面"、"小红书图片"、"做张小红书图"、"笔记配图"时使用此技能。

    1.7k GitHub stars~1.5k tokensUpdated 17 days ago
    Auto-check: notes
  • Huashu Slides

    alchaincyf/huashu-skills

    从内容到成品PPTX的端到端演示文稿制作,含AI插画生成和18种设计风格。当用户提到"做PPT"、"做幻灯片"、"演示文稿"、"Keynote"、"slides"时使用。

    1.7k GitHub stars~7.2k tokensUpdated 17 days ago
    Auto-check: notes

Questions about Huashu Proofreading

What does Huashu Proofreading do?

三遍审校降低AI检测率,让文章更有人味。当用户提到"AI味太重"、"像AI写的"、"降低AI检测率"、"审校"、"自然一些"时使用。. Huashu Proofreading is an agent skill from alchaincyf/huashu-skills.

When should I use Huashu Proofreading?

Huashu Proofreading fits situations like: tasks that involve Copy editing and proofreading.

How do I install Huashu Proofreading in Claude Code?

Run `npx skills add alchaincyf/huashu-skills --skill huashu-proofreading -a claude-code`. Or copy the skill folder (huashu-proofreading in alchaincyf/huashu-skills) into .claude/skills/huashu-proofreading in your project. Claude Code loads it when a task matches its description.

How do I install Huashu Proofreading in Codex?

Run `npx skills add alchaincyf/huashu-skills --skill huashu-proofreading -a codex`. Or copy the skill folder (huashu-proofreading in alchaincyf/huashu-skills) into .agents/skills/huashu-proofreading in your project. Codex loads it when a task matches its description.

Can I use Huashu Proofreading in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add alchaincyf/huashu-skills --skill huashu-proofreading -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/huashu-proofreading, .gemini/skills/huashu-proofreading, .github/skills/huashu-proofreading and .opencode/skills/huashu-proofreading in your project.

What does Huashu Proofreading need to run?

SKILL.md names no scripts, command-line tools or credentials: Huashu Proofreading is instructions for the agent only.

Does Huashu Proofreading access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Huashu Proofreading safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Huashu Proofreading use?

Huashu Proofreading is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Huashu Proofreading use?

About 1.2k tokens (SKILL.md is roughly 4.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Huashu Proofreading?

Skills that share tags, products or a category with Huashu Proofreading: User-Facing Text Cleanup (guillaumemeyer/watermarks-remover, 24k stars), Story Multi-Perspective Review (zenstory-ai/oh-story-claudecode, 7.4k stars), Chinese Text Humanizer (op7418/Humanizer-zh, 19k stars) and Baoyu Translate (JimLiu/baoyu-skills, 27k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Huashu Proofreading?

alchaincyf (a GitHub user) maintains it in alchaincyf/huashu-skills, which has 1,695 GitHub stars. The repository holds 21 skills in this directory. The repository was last updated on September 22, 2026.

Source: alchaincyf/huashu-skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.