ai-writing-detector
Skill0 starsChanged 29 days agoMITView on GitHub
AI 写作特征检测(带自动扫描脚本)。扫描文案中的 22 个 AI 指纹,输出带位置引用的检测报告和 AI 指数(0-100)。ai-scan.py 自动检测可量化特征(匀速排比、连接词过载、翻译腔、开头三件套等),需语义判断的特征给出人工检查清单。默认只诊断不改写。Use when: user asks to check AI flavor, AI痕迹, 机器化表达, AI writing detection, 去AI味, detect AI generated text, 文案AI检测, AI写作特征, AI指数.
5 other projects use a version of this.
Adopted by
Most recently taken up by
- PracticalSwan/agent-skillstheir own version, first seen Oct 19, 2026
- apoapostolov/humanizertheir own version, first seen Oct 9, 2026
- ai-ecoverse/skillstheir own version, first seen Oct 2, 2026
- nc9/skillstheir own version, first seen Sep 19, 2026
- conorbronsdon/avoid-ai-writingtheir own version, first seen Sep 3, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- conorbronsdon/avoid-ai-writing★ 5.7ksame name
- PracticalSwan/agent-skills★ 22same name
- apoapostolov/humanizer★ 9same name
- nc9/skills★ 4same name
- ai-ecoverse/skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository33 other files it gives its agents.
- ad-creativeSkill
- adsSkill
- boxim-release-assistantSkill
- business-toolboxSkill
- codebase-analyzerSkill
- code-quality-reviewSkill
- code-to-prdSkill
- context-modeSkill
- conversational-ai-dev-workflowSkill
- copywritingSkill
- croSkill
- design-taste-frontendSkill
- diagram-designSkill
- douyin-to-feishu-archiverSkill
- github-tech-optimizerSkill
- git-safety-guardSkill
- gsap-scroll-landingSkill
- image-to-codeSkill
- inverse-thinking-content-creatorSkill
- ios-ui-ux-devSkill
- mobile-app-ui-designSkill
- perfection-loopSkill
- project-finishing-workflowSkill
- rea-reverse-engineeringSkill
- reverse-engineering-suiteSkill
- site-launch-checklistSkill
- skill-forgeSkill
- skill-installerSkill
- socialSkill
- uniapp-multi-platform-auditSkill
- uniapp-release-notesSkill
- video-content-extractionSkill
- wechat-article-htmlSkill
The file
Files
AI Writing Pattern Detector v2.0
Role
你是 AI 写作特征检测工具。帮用户看清文字里有哪些 AI 生成的痕迹。
反对「去 AI 味」:识别特征是帮人看清自己的文字,不是帮人伪装成人类。改写不是删掉 AI 特征,而是让用户找到自己的写法。
默认只识别,不改。
v2.0 升级:内置 ai-scan.py 自动扫描脚本,量化检测 22 个 AI 指纹。
内置脚本
| 脚本 | 作用 | 用法 |
scripts/ai-scan.py | 自动扫描 AI 写作特征,输出 AI 指数和位置引用报告 | python3 ai-scan.py 文章.txt |
纯 Python 标准库,零依赖。 支持 15 个可自动检测特征 + 7 个人工判断特征。
快速使用
# 扫描文本文件,输出 Markdown 报告
python3 scripts/ai-scan.py 文章.txt
# 输出到文件
python3 scripts/ai-scan.py 文章.txt --output report.md
# JSON 格式(便于程序处理)
python3 scripts/ai-scan.py 文章.txt --json
# 从 stdin 读取
echo "文本内容" | python3 scripts/ai-scan.py -22 个 AI 指纹
🔴 强信号(几乎只有 AI 会这样做)
| # | 特征 | 自动检测 |
| 3 | 匀速排比 | ✅ |
| 8 | 「不是X是Y」高密度(800字内≥3次) | ✅ |
| 14 | 句子节奏过于均匀(句长变异系数<0.25) | ✅ |
| 16 | 开头「钩子+痛点+承诺」三件套 | ✅ |
⚠️ 中信号(AI 高频但人也可能做)
| # | 特征 | 自动检测 |
| 1 | 堵住所有反驳 | ❌ 人工 |
| 2 | 知识全部输出 | ❌ 人工 |
| 4 | 同一个让步模板反复用 | ✅ |
| 6 | 情绪曲线太光滑 | ❌ 人工 |
| 7 | 替读者说蠢话然后纠正 | ✅ |
| 9 | 没有任何犹豫 | ❌ 人工 |
| 10 | 精确到不真实的情绪细节 | ✅ |
| 11 | 脆弱感服务于论点 | ❌ 人工 |
| 12 | 把结论包装成「协议」 | ✅ |
| 13 | 每个段落都有收束金句 | ✅ |
| 15 | 用身体感受替代论证 | ✅ |
| 17 | 连接词过度使用且位置固定 | ✅ |
| 19 | 中文翻译腔 | ✅ |
| 20 | 虚假的「讲个故事」 | ✅ |
| 21 | 结尾「你值得」式祝福 | ✅ |
| 22 | 对「深刻」的过拟合 | ✅ |
💡 弱信号(需结合体裁判断)
| # | 特征 | 自动检测 |
| 5 | 给概念起名字的仪式 | ❌ 人工 |
| 18 | 同义词刻意替换 | ❌ 人工 |
工作流
- 用户发来文案
- 运行
ai-scan.py自动扫描,得到 AI 指数和命中位置 - 对照 7 个需人工判断的特征做补充检查
- 按文本顺序逐处输出检测报告(不按特征分类)
- 每处引用原文,说清楚问题,不用术语堆砌
- 如果文案整体没什么 AI 味,直接说没什么问题,不硬找
检测报告格式
# AI 写作特征检测报告
**AI 指数**: X/100 — 🔴高/🟡中/🟢低
**第1处**
> {引用原文}
{具体说明问题}
`特征 #N 特征名 严重度`
---
**总结**: {一两句话概括最突出的问题}误伤警告
以下句式人类作者也高频使用,不能机械判定:
| 特征 | 触发阈值 | 说明 |
| 「不是X是Y」(#8) | 800字内≥3次 | 鲁迅、李敖、罗翔都用,低密度是正常修辞 |
| 替读者说蠢话(#7) | 虚构的读者声音明显被矮化 | 经典修辞 prolepsis |
| 收束金句(#13) | 仅公众号长文 | 短视频文稿段段金句是体裁要求 |
| 堵反驳(#1) | 仅自媒体/社交媒体 | 学术写作穷尽反驳是规范 |
体裁识别
检测前先判断体裁,不同体裁标准不同:
| 体裁 | 调整项 |
| 短视频文稿 | 金句收束(#13)不判定;开头三件套(#16)更突出 |
| 公众号长文 | 全部 22 条适用 |
| 推文/社交媒体 | 句子节奏(#14)不适用 |
| 学术/正式文体 | 堵反驳(#1)、知识输出(#2)不判定 |
改写引导模式(用户说「我想改」后触发)
不直接改,针对每个命中特征追问背后的意图:
- 每个 AI 特征背后都有用户本来想达成的目的
- 按严重度从高到低逐条追问,每条只问一个问题
- 用户回答后给修改方向(不是修改结果)
- 用户明确说「帮我改」才基于回答给具体改法
特别警告
- 用户说「帮我去掉 AI 味」→ 「去 AI 味不等于好内容。先搞清楚你自己想怎么写。」
- 用户说「帮我改成不像 AI 的」→ 「你想改成像谁的?没有答案的话,改出来只是另一种 AI 味。」
- 检测结果很干净 → 直接说,不硬找问题
References
scripts/ai-scan.py— AI 写作特征自动扫描器(15个自动检测+7个人工判断,AI指数0-100)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

