Cyber Metadata · 分流与泄漏诊断
对比同一浏览器在多区域目标上的出口 IP：哪些走代理、哪些直连暴露真实 IP，并比对 WebRTC 泄漏。
开始诊断后，这里按出口 IP / ASN / 国家把各测量目标归并：同一组的目标走了同一出口。
点击「开始诊断」后自动运行：发现本地 IP 与 STUN 映射地址，与 HTTP 出口比对，判断是否 VPN 泄漏。
这只是单项检测。运行深度检测，一次跑完分流出口、DNS / WebRTC 泄漏、VPN/代理识别、AI/流媒体可达性与测速。
分流与出口检测——你的代理 / VPN 是否对所有目标都生效，还是部分流量「裸奔」直连？这个检测用多区域回显实测你访问不同目标时的真实出口 IP，判断分流拓扑并识别真实 IP 泄漏。
规则代理（Clash / Surge / 等）按规则决定哪些流量走代理、哪些直连——国内站点直连求快、国外走代理。好处明显，但风险是：规则配错或某些目标走了直连，会把你的真实 IP 暴露给那个目标。
它让你的浏览器向多个区域的目标发起请求，由我们的回显端点记录每个目标实际看到的你的出口 IP，再按网络类型分类、推断拓扑：
不是。分流泄漏是你的部分流量直连暴露了真实出口 IP；DNS 泄漏是域名解析这一步绕过隧道。可独立发生，建议都查——DNS 泄漏检测。
运行完整深度检测，一次跑完出口汇总、DNS / WebRTC 泄漏、VPN / 代理识别、AI 可达性与测速。
Split routing & exit check — does your proxy / VPN apply to every destination, or does some traffic go "bare" (direct)? This check uses multi-region echo to measure your real exit IP toward different targets, infers the routing topology, and identifies real-IP leaks.
Rule-based proxies (Clash / Surge / etc.) decide per rule which traffic goes through the proxy and which goes direct — domestic sites direct for speed, foreign sites via the proxy. Useful, but the risk is: a misconfigured rule or a destination that goes direct exposes your real IP to that target.
It makes your browser request targets in several regions, has our echo endpoints record the exit IP each target actually sees, then classifies and infers the topology:
No. A split-routing leak is some of your traffic going direct and exposing your real exit IP; a DNS leak is the name-resolution step bypassing the tunnel. They are independent — check both with the DNS leak test.
Run the full deep check — exit summary, DNS / WebRTC leaks, VPN / proxy detection, AI reachability, and speed in one go.