Claude Guard

Browser risk-control self-check and transparent environment diagnostics.

FAQ

What does Claude Guard check?

It checks 72 browser and network signals, including timezone, language, fonts, IP location, proxy or hosting flags, WebRTC behavior, request headers, WebGL, WebGPU, audio, permissions, automation markers, and other consistency signals.

Is it official?

No. Claude Guard is independent and does not represent Anthropic. It provides a transparent, rule-based self-check from public research and observable environment signals.

Does it work without JavaScript?

The full 72-signal browser audit needs JavaScript because browsers only expose many signals to client-side code. The documentation and utility endpoints /ip, /json, and /headers work directly over HTTP.

What endpoints should AI agents or developers use?

Use /ip for plain text public IP, /json for structured network diagnostics, /headers for request-header debugging, and /api/rules for the transparent rules database.