Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Privacy Eraser

扫描网络个人信息,提供删除方案和投诉模板,支持自动举报和定期隐私监控,助力数字身份保护。

MIT-0 · Free to use, modify, and redistribute. No attribution required.
0 · 158 · 0 current installs · 0 all-time installs
MIT-0
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill's capabilities (scanning, generating complaint templates, automating platform reports) align with its description. However, it relies on controlling a user's Mac browser/node and reading a local USER.md for identity without declaring these as required config paths or credentials. Those are reasonable technically for automatic submission, but the skill metadata does not advertise or require the sensitive resources it uses.
!
Instruction Scope
SKILL.md instructs the agent to: run nodes action=status, select a Mac node, use browser automation to navigate and click/report on third‑party sites, execute JS on pages, send private messages to platform support, take screenshots and write them to cases/, and create cron jobs for periodic scans. It also instructs reading USER.md to obtain identity. These actions reach outside a narrow 'advice/template' scope and operate on user accounts, local files, and scheduled tasks — all sensitive and potentially risky if done without explicit user confirmation.
Install Mechanism
Instruction-only (no install spec, no code files executed by installer). This is the lowest install risk: nothing is downloaded or installed automatically by the skill itself.
!
Credentials
The skill declares no env vars or credentials, yet expects access to user browser sessions (logged-in accounts on a Mac node) and to read/write local files (USER.md, cases/). Using logged‑in browser profiles to act on the user's behalf is high privilege; not declaring or requesting that explicitly is disproportionate. Templates also imply collection/submission of identity documents (IDs) in complaints, which is sensitive.
!
Persistence & Privilege
Although always:false, the skill directs the agent to create cron schedules for recurring scans and to perform autonomous automated reports via node browser control. Recurring automated actions that act through users' logged‑in sessions increase blast radius — especially combined with file writes (cases/) and messaging actions. The skill does not document safeguards (explicit confirmation prompts, audit logs, or easy disable).
What to consider before installing
Before installing or enabling this skill, consider the following: - It can control a user's logged‑in browser on a Mac node and submit reports on your behalf. Only allow this if you fully trust the environment and understand which node/profile will be used. - The runtime expects to read a local USER.md and write screenshots/cases/ files — verify where these files live and that you consent to storing identity material. Remove or redact highly sensitive docs (ID photos) unless strictly necessary. - The skill creates cron jobs for periodic scans. Ask for an option to run in 'guidance only' mode (no automated submissions) and for explicit, per‑action confirmation before any automated browser action or private messaging. - If you need automation, require a least‑privilege approach: limit node/browser access to a disposable/test profile, require interactive consent for each report, and provide an easy way to list/remove scheduled jobs and stored artifacts. - If you cannot confirm where USER.md comes from or cannot audit the node/browser sessions it will use, treat this skill as high‑risk and prefer manual guidance/templates only. - Consider testing the skill in a non‑automated mode first and review all generated complaint content and actions before allowing it to submit anything.

Like a lobster shell, security has layers — review code before you run it.

Current versionv0.2.0
Download zip
latestvk97ak46f7egsv1wx6q3xz7ezw182mfnk

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

SKILL.md

Privacy Eraser 数字身份守护

扫描 → 分析 → 给方案 → 生成投诉 → 监控


环境要求

最佳配置:Mac Mini + OpenClaw

用户的 Mac Mini 浏览器已登录各平台账号
↓
AI 通过 node 控制 Mac 上的浏览器
↓
直接自动举报,无需手动操作

检测环境

// 开始前检测
1. 调用 nodes action=status 检查可用节点
2. 如有 Mac 节点,用 browser profile=chrome target=node 控制
3. 无节点则 fallback 到指导模式

使用方式

1. 扫描模式

用户: 帮我查查网上有没有我的信息

AI 执行:
1. 从 USER.md 读取用户身份信息
2. 搜索百度/Bing(Google 被墙)
3. 只报告有问题的内容
4. 给出最有效的处理方案

2. 定向模式

用户: 帮我删这个 [URL]

AI 执行:
1. 识别平台
2. 查平台潜规则
3. 生成投诉材料
4. 告诉用户最快的提交方式

3. 监控模式

用户: 帮我监控隐私

AI 执行:
1. 设置 cron 每周扫描一次
2. 发现新内容主动通知

4. 自动举报流程(需要 Mac 节点)

用户: 帮我删这个 [URL]

AI 执行:
1. 检测 Mac 节点: nodes action=status
2. 有节点 → 用 browser target=node profile=chrome
3. 导航到目标 URL
4. 识别平台,执行对应自动化:

百度文库:
  - 滚动到版权说明区域
  - 点击"举报"
  - 选择"侵犯隐私"
  - 填写说明
  - 提交

知乎:
  - 点击"举报"按钮
  - 选择"侵犯个人隐私"  
  - 提交

微博:
  - 点击"..."菜单
  - 选择"举报"
  - 选择"泄露隐私"
  - 提交

5. 截图存档
6. 记录到 cases/ 目录

自动化脚本

百度文库举报

1. browser navigate 到文档页
2. 执行 JS: document.querySelector('.report-click-item').click()
3. 等待举报弹窗
4. 选择举报类型(侵犯隐私)
5. 填写说明:"该文档包含本人真实姓名等个人信息,未经本人授权公开,请予删除"
6. 点击提交
7. 截图存档

知乎举报

1. browser navigate 到目标内容页
2. 点击"举报"按钮
3. 选择"侵犯个人隐私"
4. 补充说明
5. 提交
6. 截图存档

微博举报

1. browser navigate 到微博详情页
2. 点击右上角"..."
3. 选择"举报"
4. 选择"泄露个人隐私"
5. 提交
6. 同时私信 @微博客服 加速处理
7. 截图存档

Google 搜索结果移除

1. browser navigate 到 google.com/webmasters/tools/legal-removal-request
2. 选择"个人信息"
3. 填写 URL 和说明
4. 提交
5. 截图存档
(无需登录,但需要翻墙)

平台潜规则(实测有效)

百度文库

❌ 别用: tousu.baidu.com(已废弃,404)
✅ 最快: 手机APP → 右上角"..." → 举报 → 侵犯隐私
✅ 备选: 电脑端文档页底部"版权说明"→ 举报
⏱️ 处理时间: 48小时
💡 技巧: 选"侵犯隐私"比"其他"快

百度快照

✅ 入口: 搜索结果右下角"快照"→"投诉快照"
⚠️ 前提: 原页面必须已删除,否则快照删不掉
⏱️ 处理时间: 1-3天

百度搜索结果

✅ 入口: help.baidu.com/webmaster/add
✅ 选择: 个人信息保护
⏱️ 处理时间: 3-7天

知乎

❌ 别用: 普通"举报"按钮(慢)
✅ 最快: zhihu.com/terms/complaint → 个人信息保护
💡 技巧: 附上身份证明处理更快
⏱️ 处理时间: 3-5天
⚠️ 潜规则: 大V内容难删,匿名回答容易删

微博

✅ 方法: 举报 + 同时私信 @微博客服(双管齐下)
💡 技巧: 只举报不私信,处理很慢
⏱️ 处理时间: 2-7天
⚠️ 潜规则: 蓝V发的难删,转发多的几乎删不掉

政府网站

⚠️ 现实: 基本不会删
✅ 替代方案: 删搜索引擎快照,让它搜不到
📝 如果非要试: 联系信息公开办公室,引用《个人信息保护法》第47条

Google

✅ 入口: google.com/webmasters/tools/legal-removal-request
✅ 选择: 个人信息 → 具体类型
⏱️ 处理时间: 1-2周
💡 国内访问需翻墙

微信公众号

✅ APP内: 文章底部 → "投诉" → 选择"侵犯隐私"
✅ PC端: 文章页右上角 "..." → 投诉
✅ 专用通道: https://weixin110.qq.com (隐私侵权)
⏱️ 处理时间: 3-7天(普通),1-3天(有证据)
💡 技巧: 律师函/公证书 → 快很多
💡 技巧: 侵权投诉比"其他"处理优先级高
⚠️ 潜规则: 普通用户投诉排队慢,已过原创的文章更难删

小红书

✅ APP内: 笔记右上角 "..." → 举报 → "涉及隐私"
✅ 创作者中心: https://creator.xiaohongshu.com
✅ 邮箱: jubao@xiaohongshu.com(侵权投诉)
⏱️ 处理时间: 24-72小时
💡 技巧: "人身攻击"比"隐私泄露"更快处理
💡 技巧: 多人举报会加速
⚠️ 潜规则: KOL/大V 的内容基本删不掉,素人笔记举报成功率高

抖音

✅ APP内: 视频右下角 "..." → 举报 → "侵犯隐私"
✅ 创作者中心: https://creator.douyin.com
✅ 侵权投诉: https://www.douyin.com/draft/copyright/complaint
⏱️ 处理时间: 1-7天
💡 技巧: "未经同意拍摄"成功率较高
💡 技巧: 私信博主协商删除是最快路径
⚠️ 潜规则: 视频比图文难删,播放量高的内容几乎删不掉

模板

中国平台

  • templates/baidu_removal.md - 百度系
  • templates/zhihu_report.md - 知乎
  • templates/weibo_report.md - 微博
  • templates/weixin_report.md - 微信公众号
  • templates/cyberspace_report.md - 网信办举报

国际平台

  • templates/google_removal.md - Google
  • templates/gdpr_request.md - GDPR(欧洲)
  • templates/dmca_takedown.md - DMCA(美国)

通用

  • templates/generic_privacy.md - 通用隐私投诉
  • templates/legal_warning_cn.md - 律师函模板

优先级判断

内容类型紧急度处理方式
手机号/身份证🔴 立即直接举报,不协商
住址/工作单位🟡 24h内举报
真名+学校🟡 48h内举报或协商
仅姓名提及🟢 可选看情况,可忽略

监控设置

用户要求监控时,创建 cron:

{
  "schedule": { "kind": "cron", "expr": "0 9 * * 1", "tz": "Asia/Shanghai" },
  "payload": { "kind": "systemEvent", "text": "隐私监控: 搜索用户姓名,检查是否有新的个人信息泄露,如有则通知用户" },
  "sessionTarget": "main"
}

节点浏览器控制

检查可用节点

nodes action=status
→ 查看是否有 Mac 节点在线

使用节点浏览器

browser action=open target=node profile=chrome targetUrl="https://..."
browser action=snapshot target=node
browser action=act target=node request={...}

为什么用节点浏览器?

1. 用户的 Mac 浏览器已登录各平台
2. 不需要用户手动提供 Cookie
3. 完全自动化,用户无感知
4. 支持所有需要登录的平台

参考资料

  • references/platforms.md - 各平台举报入口汇总
  • references/legal_basis.md - 法律依据(个保法、网安法等)

Files

15 total
Select a file
Select a file to preview.

Comments

Loading comments…