Casino Scam Detection
The Casino detection engine identifies fake casino promotions that use celebrity names (MrBeast, Elon Musk, Ronaldo, Taylor Swift, etc.) to lure victims.How It Works
The engine uses a two-stage analysis pipeline:1. Text Analysis
Extracts text from the message content, attachment filenames, and embed metadata. It checks for four categories of terms:| Category | Examples | Score Contribution |
|---|---|---|
| Celebrity | mrbeast, elon, ronaldo, taylor swift, drake, ksi, trump | +1–3 per term |
| Casino | casino, slots, bonus, promo, wager, rakeback | +1–4 per term |
| Crypto | wallet, usdt, bnb, withdraw, deposit, erc20 | +1–4 per term |
| Pressure | claim, giveaway, limited, congratulations, fastest | +1–3 per term |
2. Image Analysis
If text analysis doesn’t reach the threshold, the bot downloads and analyzes attached images using pixel-level feature detection:- Dark neon UI detection — Measures dark pixel ratio, blue/purple balance, and color saturation
- Green button detection — Identifies green “claim” or “bonus” action buttons
- Text-heavy screenshot detection — Measures bright text pixels and edge density
- Reference matching — Compares image features against known casino scam images
Configuration
- Threshold (default: 5) — Minimum score required to trigger detection. Lower values are more sensitive.
- Actions — Delete, warn, timeout, kick, ban (configurable order)
- Alerts — Toggle alert embeds, set channel, configure ping roles
Message Triggers
The casino engine only processes messages that contain image attachments. Text-only messages are not analyzed by this engine (other engines cover text scams).Debug Mode
SetCASINO_SCAM_DEBUG=true in the .env file to enable verbose logging of image analysis results, including feature ratios and match distances.