AntiFraud SDK
Behavioral risk signal for forms: JS client plus FastAPI scoring. Sends aggregates only (no field values, key identities or coordinates); signed result token your backend verifies; fails open. Weights are not calibrated on real users.
Production-oriented White-Label modules for P2P, iGaming and fintech.
Behavioral risk signal for forms: JS client plus FastAPI scoring. Sends aggregates only (no field values, key identities or coordinates); signed result token your backend verifies; fails open. Weights are not calibrated on real users.
Receipt and transfer-screenshot check (FastAPI): ELA, EXIF, Tesseract OCR with an expected-amount cross-check, duplicate detection. Returns a transparent score breakdown. A signal for manual review, not proof of forgery; ELA only catches a clearly different compression history; weights are not calibrated on real receipts.
UI widget for 60-second micro-rounds: sandboxed iframe plus postMessage protocol v1, countdown to the host's close time, two-step confirmation, lock command for responsible gaming. UI only: it sends a bet intent to your system and does not take bets, hold money or settle. Needs a licensed operator.
Interactive hotspots over video from timed ID3 metadata (hls.js or native text tracks) or manual push: own ID3v2 decoder, correct geometry for object-fit contain/cover, keyboard access. UI only: it reports clicks to your system and executes nothing. Not tested on a real HLS stream or Safari/Firefox.
Fiat-to-USDT order gateway on FastAPI and PostgreSQL: quotes with fee, orders, PSP webhook confirmation, USDT payouts (TRC20/ERC20), OTP verification tiers with limits, self-exclusion, signed operator webhooks. You connect your own rate feed, PSP and wallet keys.
Email, phone, IP, VPN/Tor and disposable-domain risk scoring, 0–100. Stateless; IP lists are downloaded by a bundled updater.
Email or domain breach checks via public sources (XposedOrNot, Hudson Rock; HaveIBeenPwned with a paid key): risk score and recommendations.
Screening of Russian legal entities and sole proprietors by INN or name: registry status, courts, bankruptcy, sanctions. Sources and limits are in the README.
Voice deepfake-risk signal: 0–100 score and verdict. Heuristics plus an optional neural model (about 400 MB). Not calibrated on real recordings.
Contract risk check: rules plus an LLM (Gemini API), with quotes, missing sections and a risk score. Contract text is sent to Gemini. Not legal advice.
Single-file HTML dashboard: module status, KPIs, live feed, audit log. Ships with demo data; connects to a real API through a documented contract.
Receipt Shield + AntiFraud + LeakGuard
AntiFraud + Micro-Betting + Overlay + Gateway
All 11 modules + demo panel