Engineering notes on browser fingerprints, proxy compatibility and reproducible detector checks.
Font lists can provide 9.45 bits of entropy and distinguish 35% of measured devices. This guide explains the signals inside a fingerprint, our baseline measurements and which surfaces should remain unmodified.
read the analysisAudio variation is still tied to the site origin rather than the profile. We show where this appears in NimbusX, how to account for it today and what an engine-level fix requires.
read the analysisChromium cannot authenticate to SOCKS5 proxies and some proxies stall on CONNECT requests with a Host header. Here is how the built-in relay handles both cases and why NimbusX checks location before launch.
read the analysisSeven detectors, rated 100 out of 100 on engine 154.0.8037.58. What everyone checks, why TLS cannot be faked from JavaScript and why pixelscan is calculated manually.
read the analysisAn Android profile on a desktop computer received Iphey's Trustworthy verdict. We document the Pixel 7 chipset mismatch that led us to remove a hand-maintained device corpus, plus six tests that show the remaining desktop limits.
read the analysisThe dashboard, CLI, API and MCP share one service layer. The guide covers REST profile creation, safe automation and a working MCP configuration.
read the analysisA pool of 100 to 300 common device configurations is usually enough. We explain why uncommon hardware is easier to spot and review three inconsistencies caught by the validator before release.
read the analysisTake your fingerprint and look next to the profile NimbusX. Or download and try it yourself.
Open check