CodeCloner 现在有两个本事:
模块一:抄代码(原「抄袭者」)
你看到一个开源项目,心想:"这东西真好,要是叫我的名字就好了。"
CodeCloner 的抄袭者模式会自动帮你:
- 🔍 查清项目能不能抄、怎么抄(许可证检查)
- 📋 大白话告诉你项目是什么、有什么、要改哪里
- ✏️ 按你的要求改名字、换品牌、删功能、加功能
- ✅ 验证改完的东西能跑
- 📦 打包交付,推到你的仓库
模块二:抄网页(新增)
你看到一个网站设计得好看,心想:"这个页面要是能离线打开就好了。"
CodeCloner 的抄网页模式会自动帮你:
- 🌐 用浏览器加载目标网址
- 🔒 绕过 CDN 403,直接抓取所有 CSS 和图片
- 📦 把 CSS 内联、图片 base64,生成一个单文件 HTML
- 🖼️ 截屏留档
- 📤 交付给你,浏览器直接打开,完全离线可用
全程不用写一行代码。你只管丢链接,AI 帮你干活。
CodeCloner 不绑定任何平台——任何能读 markdown 指令的 AI Agent 都能用:
| 平台 | 怎么用 |
|---|---|
| 🟣 Claude Code | 原生 Skill 格式,直接放到 ~/.claude/skills/ 或项目 .claude/skills/ |
| 🟢 OpenAI Codex | 把 SKILL.md 内容贴进项目根目录的 AGENTS.md 或作为 system prompt |
| 🟡 OpenClaw | 作为 Agent 指令文件导入,或贴进对话开头 |
| 🔵 Hermes Agent | 作为 skill/instruction 文件加载,或贴进 system prompt |
| 🔘 其他 AI Agent | 只要能读 markdown 规则,就能用 |
Claude Code
# 全局安装(所有项目都能用)
git clone https://github.com/Eamon3949/CodeCloner.git ~/.claude/skills/CodeCloner
# 项目级安装(只在当前项目能用)
git clone https://github.com/Eamon3949/CodeCloner.git .claude/skills/CodeCloner装完后在任何对话里输入以下关键词即可触发:
| 你想干什么 | 说这个 |
|---|---|
| 抄代码仓库(改名换姓) | [激活抄袭者] https://github.com/xxx/yyy |
| 抄网页(离线保存) | [抄网页] https://example.com |
| 自动判断 | 直接丢链接,系统自动判断走哪个模块 |
抄网页需要额外装 Playwright 浏览器内核(AI Agent 会自动帮你装,不用操心)
OpenAI Codex
cp SKILL.md /your/project/AGENTS.mdOpenClaw / Hermes Agent / 其他平台
把 SKILL.md 的内容复制粘贴到对话开头,或作为 system prompt / instruction 文件导入。
| 特点 | 说明 |
|---|---|
| 🛡️ 许可证把关 | 抄代码时自动检查 MIT / Apache / GPL 等许可证,告诉你能不能抄、要遵守什么 |
| 🔤 命名全覆盖 | 抄代码时 PascalCase、camelCase、kebab-case、snake_case、常量、URL 全部替换,一个不漏 |
| 🧹 品牌清洗 | 抄代码时按 P0 / P1 / P2 优先级逐项清洗,内置 0-100 分干净度评分系统 |
| 🔒 CDN 破解 | 抄网页时 CDN 403 防盗链也没用,浏览器能看 = 就能拿到资源 |
| 🎯 1:1 复刻 | 抄网页时 CSS 全部内联 + 图片 base64,与原站几乎一致 |
| 💬 大白话汇报 | 全程用你听得懂的话解释,不懂代码也能做决策 |
| 🔇 静音模式 | 加上 --auto 全自动跑,只在你需要的时候才打断你 |
| 📖 参考文档 | 附带 7 份详细参考(工作流、品牌清单、命名变体、评分系统、页面克隆等) |
你需要什么?
- 抄代码:一个 Git 平台账号 + Git + AI Agent 工具
- 抄网页:AI Agent 工具 + Playwright(会自动安装)
- 不需要懂代码
抄代码:三步搞定
第 1 步 ── 发送项目链接
你:https://github.com/某人/某个项目 [激活抄袭者]
第 2 步 ── 看报告,提需求
AI:📋 可行性报告:MIT 协议 ✅ 可以抄,约 120 个文件,预计分 2-3 批改完
你:项目叫 MyApp,去掉社区功能
第 3 步 ── 等交付
AI:✅ 改完了!已推送到你的仓库
抄网页:也是三步搞定
第 1 步 ── 发送网址
你:[抄网页] https://example.com
第 2 步 ── 等克隆
AI:🌐 正在加载页面... ✅ CSS 17 个(734KB)全部内联 ✅ 图片 16 张全部 base64
第 3 步 ── 打开看效果
AI:✅ 已生成 full_clone.html(9.2MB),浏览器打开直接看
静音模式
你:https://github.com/某人/某个项目 [激活抄袭者] --auto
AI:(安静地干活,只在出问题时才打断你)
┌─────────────┐
│ 第 0 步 │ 🛡️ 可行性评估 → 许可证检查 / 规模评估 / 技术栈检查
└──────┬──────┘
│ ✅ 可行
▼
┌─────────────┐
│ 第 1 步 │ 🔍 项目拆解 → 克隆仓库 / 分析结构 / 识别品牌要素
└──────┬──────┘
▼
┌─────────────┐
│ 第 2 步 │ 📋 大白话汇报 → 一句话说清楚 / 核心亮点 / 问你想改什么
└──────┬──────┘
▼
┌─────────────┐
│ 第 3 步 │ ✏️ 交互式重构 → 品牌清洗 / 功能裁剪 / 功能增强
└──────┬──────┘
▼
┌─────────────┐
│ 第 4 步 │ ✅ 冒烟验证 → 编译检查 / 安全扫描 / 运行验证 / 干净度评分
└──────┬──────┘
▼
┌─────────────┐
│ 第 5 步 │ 📦 极简交付 → Dockerfile / 一键启动脚本 / 推送仓库
└─────────────┘
| 维度 | 满分 | 扣分规则 | 说明 |
|---|---|---|---|
| P0 核心替换 | 40 | 每处残留扣 5 分 | 项目名没换完 = 侵权风险 |
| P0 包名/import | 30 | 每处残留扣 10 分 | import 路径没改 = 项目跑不起来 |
| P1 文档配置 | 15 | 每处残留扣 3 分 | README、package.json 里的旧名 |
| P2 辅助信息 | 15 | 每处残留扣 1 分 | CHANGELOG 中的历史记录 |
| 分数 | 等级 | 结论 |
|---|---|---|
| 95-100 | 🟢 A | 可以放心推送 |
| 80-94 | 🟡 B | 可以交付,P2 有合理残留 |
| 60-79 | 🟠 C | 不建议交付,继续洗 |
| < 60 | 🔴 D | 禁止交付 |
| 许可证 | 能不能抄 | 归因要求 |
|---|---|---|
| MIT | ✅ 能 | 保留版权声明 |
| Apache-2.0 | ✅ 能 | 保留版权 + NOTICE |
| BSD | ✅ 能 | 保留版权声明 |
| GPL | 你的项目也必须开源 | |
| AGPL | 网络服务也要开源 | |
| Unlicense | ✅ 能 | 无要求 |
| 无许可证 | 🚫 不能 | 法律上 = 保留所有权利 |
- 本 Skill 产生的项目基于原项目的开源许可证,请遵守原作者的许可条款
- GPL / AGPL 项目抄完后必须同样开源,这是法律要求,不是建议
- 不要删除原作者的 LICENSE 文件,而是替换版权声明中的作者名和项目名
- 商业使用请确保获得原项目作者授权
CodeCloner/
├── SKILL.md ← 技能入口(路由规则 + 6 步工作流)
├── VERSION ← 版本号
├── LICENSE ← MIT 许可证
├── CHANGELOG.md ← 变更日志
├── logo.svg ← 项目 Logo
├── README.md ← 你正在看的这个文件
├── .gitignore
│
├── references/ ← 抄代码模块参考文档
│ ├── workflow.md ← 详细工作流(含技术栈决策树)
│ ├── branding-checklist.md ← 品牌清洗检查清单
│ ├── naming-variants.md ← 命名变体映射规则
│ └── cleanliness-score.md ← 干净度评分系统
│
└── cloner-web/ ← 抄网页模块
├── SKILL.md ← 网页克隆专属规则
├── scripts/
│ └── clone_intercept.py ← Playwright 拦截脚本
└── references/
└── page-cloning-notes.md ← 页面克隆技术笔记
| 文档 | 一句话说明 |
|---|---|
| workflow.md | 抄代码:每一步具体干什么、怎么干、遇到问题怎么办 |
| branding-checklist.md | 抄代码:品牌清洗 P0/P1/P2 优先级清单 + grep 模板 |
| naming-variants.md | 抄代码:8 种命名变体全覆盖规则 |
| cleanliness-score.md | 抄代码:0-100 分干净度评分系统 |
| cloner-web/SKILL.md | 抄网页:完整工作流和操作指南 |
| cloner-web/scripts/clone_intercept.py | 抄网页:Playwright 响应拦截脚本 |
| cloner-web/references/page-cloning-notes.md | 抄网页:技术原理和已知陷坑 |
欢迎贡献!你可以:
- 提 Bug — 在 Issues 里描述问题
- 提建议 — 在 Issues 里标记为
enhancement - 提 PR — Fork 仓库,改完后提 Pull Request
- 加平台 — 如果你用过的 AI Agent 平台不在列表里,告诉我们怎么适配
贡献前请确保:
- 改动不影响 SKILL.md 的通用性(不要加平台特定的硬编码)
- 新加的参考文档放在
references/下 - 更新 CHANGELOG.md
You found an amazing open-source project. You love it. But it's not yours. Or you saw a beautiful website that you wish you could save offline.
CodeCloner now does both.
Module 1: Clone Code (the original "Cloner")
Clone any open-source repo, rebrand it, and ship it as your own. Just say [activate cloner] + repo URL.
- 🔍 License check — can you legally clone this?
- 📋 Plain-language report — what does it do, what needs changing
- ✏️ Interactive rebranding — rename, recolor, remove, add
- ✅ Smoke test — does it still run?
- 📦 Ship — push to your own repo
Module 2: Clone Web (new!)
Clone any website into a self-contained single HTML file. Say [clone web] + URL.
- 🌐 Browser loads the page (Playwright)
- 🔒 Intercepts all CSS & images inside the browser — bypasses CDN 403
- 📦 Inlines CSS + base64 images → single standalone HTML
- 🖼️ Screenshot for reference
- 📤 Delivers — open in browser, works offline
Zero coding. Just drop a link, let the AI do the work.
CodeCloner is not locked to any single platform — any AI agent that reads markdown instructions can use it:
| Platform | How to use |
|---|---|
| 🟣 Claude Code | Native Skill format — drop into ~/.claude/skills/ or project .claude/skills/ |
| 🟢 OpenAI Codex | Paste SKILL.md content into your project's AGENTS.md or use as system prompt |
| 🟡 OpenClaw | Import as agent instruction file, or paste at the start of a conversation |
| 🔵 Hermes Agent | Load as skill/instruction file, or paste into system prompt |
| 🔘 Other AI Agents | If it reads markdown rules, it works |
Claude Code
# Global install (available in all projects)
git clone https://github.com/Eamon3949/CodeCloner.git ~/.claude/skills/CodeCloner
# Project-level install (current project only)
git clone https://github.com/Eamon3949/CodeCloner.git .claude/skills/CodeClonerThen type [activate cloner] + a project URL in any conversation to trigger it.
OpenAI Codex
cp SKILL.md /your/project/AGENTS.mdOpenClaw / Hermes Agent / Other platforms
Copy-paste the content of SKILL.md at the start of your conversation, or import it as a system prompt / instruction file. It's universal markdown — no special adaptation needed.
| Feature | Description |
|---|---|
| 🛡️ License Guard | Automatically checks MIT / Apache / GPL licenses — tells you if you can clone it and what you must comply with |
| 🔤 Full Name Coverage | Not just a find-and-replace — handles PascalCase, camelCase, kebab-case, snake_case, SCREAMING_SNAKE, constants, and URLs. All 8 variants, zero misses |
| 🧹 Brand Cleanup | P0 / P1 / P2 priority-based cleanup with a built-in 0-100 Cleanliness Score system |
| 💬 Plain Language Reports | Explains everything in terms anyone can understand — no coding knowledge needed |
| 🔇 Silent Mode | Add --auto for fully hands-off operation. Only interrupts you when it really needs to |
| 📖 Reference Docs | 4 detailed references included (workflow, branding checklist, naming variants, scoring system) |
What you need:
- A GitHub / GitLab / Bitbucket account
- Git installed
- Claude Code or Codex or another AI Agent tool installed
- For auto-push: GitHub CLI (gh) installed
- No coding knowledge required
Three steps:
Step 1 ── Send the project URL
You: https://github.com/someone/some-project [activate cloner]
Step 2 ── Read the report, give instructions
AI: 📋 Feasibility: MIT license ✅, ~120 files, 2-3 batches estimated
You: Name it MyApp, remove the community features
Step 3 ── Wait for delivery
AI: ✅ Done! Pushed to your repository
Silent mode
Don't want step-by-step questions? Add --auto — the AI runs on defaults and only interrupts if something goes wrong:
You: https://github.com/someone/some-project [activate cloner] --auto
AI: (quietly working, only asks when something needs your decision)
┌─────────────┐
│ Step 0 │ 🛡️ Feasibility → License check / Scale assessment / Tech stack
└──────┬──────┘
│ ✅ viable
▼
┌─────────────┐
│ Step 1 │ 🔍 Analysis → Clone repo / Map structure / Identify brand elements
└──────┬──────┘
▼
┌─────────────┐
│ Step 2 │ 📋 Briefing → One-liner summary / Key highlights / Ask what to change
└──────┬──────┘
▼
┌─────────────┐
│ Step 3 │ ✏️ Rebranding → Brand cleanup / Feature trimming / Feature additions
└──────┬──────┘
▼
┌─────────────┐
│ Step 4 │ ✅ Verification → Compile check / Security scan / Runtime test / Cleanliness score
└──────┬──────┘
▼
┌─────────────┐
│ Step 5 │ 📦 Delivery → Dockerfile / One-click scripts / Push to repo
└─────────────┘
How clean is "clean enough"? We quantify it:
| Dimension | Max | Deduction | Meaning |
|---|---|---|---|
| P0 Core Names | 40 | -5 per miss | Unreplaced project names = legal risk |
| P0 Imports | 30 | -10 per miss | Unreplaced imports = project won't run |
| P1 Docs/Config | 15 | -3 per miss | Old names in README, package.json, etc. |
| P2 Misc | 15 | -1 per miss | Historical entries in CHANGELOG |
| Score | Grade | Verdict |
|---|---|---|
| 95-100 | 🟢 A | Ship it |
| 80-94 | 🟡 B | Deliverable, minor P2 residue |
| 60-79 | 🟠 C | Not recommended, keep cleaning |
| < 60 | 🔴 D | Delivery blocked |
| License | Can clone? | Attribution required |
|---|---|---|
| MIT | ✅ Yes | Keep copyright notice |
| Apache-2.0 | ✅ Yes | Keep copyright + NOTICE |
| BSD | ✅ Yes | Keep copyright notice |
| GPL | Your project must also be open-source | |
| AGPL | Network services must also be open-source | |
| Unlicense | ✅ Yes | None |
| No license | 🚫 No | Legally = all rights reserved |
- Projects produced by this Skill are subject to the original project's open-source license — comply with the original author's terms
- GPL / AGPL projects must remain open-source after cloning — this is a legal requirement, not a suggestion
- Do not delete the original author's LICENSE file — replace the author name and project name in the copyright notice
- For commercial use, ensure you have the original author's authorization
CodeCloner/
├── SKILL.md ← Skill entry point (6-step workflow rules)
├── VERSION ← Version number
├── LICENSE ← MIT License
├── CHANGELOG.md ← Change log
├── logo.svg ← Project logo
├── README.md ← This file
├── .gitignore
└── references/
├── workflow.md ← Detailed workflow (with tech stack decision tree)
├── branding-checklist.md ← Brand cleanup checklist (P0/P1/P2 priorities + grep templates)
├── naming-variants.md ← Naming variant mapping rules (8 variants covered)
└── cleanliness-score.md ← Cleanliness scoring system (0-100)
| Document | One-liner |
|---|---|
| workflow.md | What to do at each step, how to do it, and what to do when things go wrong |
| branding-checklist.md | Follow this checklist when renaming — P0 items are mandatory |
| naming-variants.md | 8 naming variant forms — replace every single one |
| cleanliness-score.md | 0-100 scoring system — 90+ to ship |
Contributions welcome! You can:
- Report bugs — Describe the issue in Issues
- Suggest features — Tag as
enhancementin Issues - Submit PRs — Fork the repo, make changes, and open a Pull Request
- Add platforms — If your AI Agent platform isn't listed, tell us how to adapt it
Before contributing:
- Keep SKILL.md platform-agnostic (no platform-specific hardcoding)
- Put new reference docs under
references/ - Update CHANGELOG.md
This project is open-sourced under the MIT License.
本项目采用 MIT License 开源。
Made with 🔥 by Eamon3949