Fingerprints

A claim farm can rent a thousand home IPs and use each one once. It's much harder to run a thousand different machines. Fingerprints looks at the machine.

What it collects

When a visitor passes through the gym, a small script records how their browser draws, sounds and counts: canvas, WebGL, audio, fonts and maths, plus the GPU, screen, cores, memory, timezone and languages. The results are hashed, sealed and signed for that one session, so a farm can't replay a clean fingerprint from somewhere else.

What it catches

• One image, many IPs: the same canvas and WebGL hashes behind hundreds of "different" visitors, and how often each one comes back.
• Browsers that lie: a user agent that claims to be Chrome on Windows while its client hints say otherwise.
• Automation: the webdriver flag, and hardware that doesn't fit the device it claims to be.

Where it shows up

Fingerprints show up beside clients in Activity, tied to each one's IP and TLS fingerprint, and what they find feeds the spoofed and outdated client traits your rules already match on, so a confirmed farm can be sent straight to the gym next time. The Fingerprints page filters by hash, GPU, platform, time zone, window size, automation and mismatches, and groups by TLS fingerprint, IP or client.