go-forge · reese84 device harvester

Open this page in the real browser / device you want to impersonate (genuine Chrome / Firefox / Safari / mobile). It runs the real reese84 sensor locally with the network neutralized — nothing is posted to Imperva — and captures everything the offline V8 forge needs to spoof this device, including the full audio fingerprint render. Profiles save to tools/go-forge/profiles/.

1 · sensor source Live reese84 sensor URL — paste the target's reese84 script URL. The server fetches it fresh per capture (the sensor rotates ~every 10 min). Leave blank to use the repo's bundled script.js if one is present; required when running from a dist tree that ships no sensor.

2 · label this device Device / browser label (becomes the profile filename)

2b · TLS fingerprint echo Fingerprint-echo URL (JA3/JA4/Akamai-H2) — default /harvest/tls is THIS harvester's own same-origin endpoint (no CORS, no 3rd party): it fingerprints the device's real ClientHello (JA3/JA4/peetprint) + captured HTTP/2 frames (Akamai). Requires the harvester to run with -tls (an https origin). Saved into the profile for later replay in custom libs. Captured every harvest (fail-soft). Blank = skip. Override with an external echo (e.g. https://tls.peet.ws/api/all) if you prefer.

3 · capture

capture offline self-test fixture — freezes this page's RNG + clock to the forge's deterministic config (Math.random=det seed 7, crypto=det 2654435761, Date=epoch 1780360000000, performance.now=monotone) and captures the sensor's OWN emitted p + envelope so /solver can byte-verify the forge against it. Saves as <label>.selftest.json. Leave OFF for a normal live-spoof profile.

Idle. Click “Capture this device”.

4 · log

User-Agent
browser / version
health
interrogate
audio buffer
audio reductions
canvas toDataURL
fingerprint PNG
measureText (fonts)
getImageData
WebGL
window for-in / ownTypes
neutralized requests
self-test fixture