Communitygithub.com

Qiaomu AI Access

AI access environment signal check and compliant privacy-hygiene skill

What is Qiaomu AI Access?

Qiaomu AI Access is a developer skill for AI agent platforms like Claude Code, Cursor, and Codex. It helps detect runtime, browser, and network signals that may indicate a user's AI access environment—such as language, timezone, fonts, emoji, and the `isChinaUser` result—without bypassing service restrictions or misrepresenting identity. The skill runs a safe detector using the npm package `is-china-user` in Node.js and optionally in a headless Playwright browser context, producing a detailed report with coverage tables and consent prompts for compliant privacy hygiene. It explicitly refuses requests for VPN/proxy purchase, IP masking, account-region laundering, CAPTCHA bypass, or other evasion techniques, focusing instead on transparent signal reporting and non-destructive steps to reduce unnecessary regional signals in prompts and configurations. This tool is useful for developers who need to understand how AI services might perceive their environment, while staying within legal and platform boundaries.

Works with~Claude Code~Codex CLI~Cursor
npx skills add joeseesun/qiaomu-ai-access

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

What does Qiaomu AI Access do?

Detect first, ask second, stay inside legal and platform boundaries.

Router Rules

  • Use this skill when the user's goal is environment-signal transparency before using AI websites, APIs, or top AI models.
  • Use it when the user asks to identify "中国用户特征", "China user signals", "AI access signals", browser language/timezone/font/emoji signals, or isChinaUser results.
  • If the user asks directly for VPN/proxy purchase, IP masking, account-region laundering, KYC/payment misrepresentation, CAPTCHA bypass, device-fingerprint evasion, or sanctions/geofence bypass, do not run an evasion workflow. Briefly refuse that part and offer the safe detection and privacy-hygiene workflow.
  • Do not treat nationality, ethnicity, residence, or legal eligibility as something this skill can determine. It only reports runtime signals.

Workflow

  1. State the boundary in one sentence: this skill checks signals and can suggest compliant privacy hygiene, but it does not help bypass service restrictions or misrepresent identity/location.

  2. Run the detector from the repository root after dependencies are installed:

    npm install
    npm run detect:browser -- --output reports/latest-ai-access-check.md
    

    If the user explicitly agrees to remote network probes, run:

    npm run detect:full -- --output reports/latest-ai-access-check.md
    
  3. Show the user the high-level status, coverage level, and signal table. Treat runtime, browser, and network coverage separately; do not overstate skipped or unavailable layers.

  4. Ask this exact confirmation before any follow-up hygiene advice:

    是否继续做合规隐私卫生检查?我可以帮你减少 prompt、浏览器偏好和工作区说明里不必要的地域信号;不会帮助绕过平台地域限制、伪装 IP/身份、规避风控或违反服务条款。
    
  5. If the user says yes, provide only non-destructive, compliant steps:

    • prefer official AI services, models, regions, and payment paths available to the user;
    • use English prompts/UI preferences when the goal is English model behavior or documentation quality;
    • keep a separate browser profile for AI work to reduce cookie/history/permission mixing;
    • remove unnecessary location claims from prompts, docs, and profile snippets when they are not relevant;
    • explain that changing system language/timezone to misrepresent location or bypass controls is outside this skill.
  6. If the user says no or seems unsure, stop after the report and offer to save it for later.

Output Contract

  • reports/latest-ai-access-check.md when the detector is run with --output.
  • A concise status: china-signals-detected, no-china-signal-detected, or inconclusive.
  • A coverage table for runtime, browser, and network.
  • A signal table with source, value, and notes.
  • The required consent prompt.
  • Clear unavailable/missing-evidence labels for browser-only or network-only signals not measured in the current run.

Runtime Notes

  • The first-step detector imports the npm package is-china-user.
  • Node.js can reliably inspect timezone and limited runtime locale data.
  • Browser coverage uses a temporary Playwright Chromium/headless-shell context, injects the installed is-china-user package bundle into that page, and does not read a personal browser profile.
  • DOM/canvas-based emoji and font checks should be interpreted from the browser layer, not from Node-only runs.
  • Optional network probing is off by default because it creates real remote requests and is noisy under proxies, extensions, DNS failures, or offline conditions.
  • The upstream package currently has no declared license in GitHub/npm metadata, so this skill depends on it without vendoring its source.

Validation

Run before publishing changes:

npm test
npm run smoke:browser
npm run validate:skill
npm run eval:trigger
npm run export:ir
npm run secret:scan

Release Boundary

  • Public claims must match the validation output.
  • Do not publish secrets, local browser profiles, cookies, .env files, network credentials, or screenshots containing account data.
  • Keep the README Chinese-first bilingual for Qiaomu-owned public releases.

Related Skills