Antiscammer (AS) Antiscammer is a multi-tenant Scam Intelligence Platform for Discord — a SaaS that protects servers from scammers, fraud links, and malicious auto-responders. Originally a Discord moderation bot, it's being repositioned (V3) into a broader threat-intelligence product.
What it does AI scam detection — a multi-agent pipeline (Evidence → Risk → AI → aggregation) scans messages and images, scores confidence, and files structured reports. Threat intelligence — maintains a shared blocklist (~761k scam URLs) imported from threat feeds, plus staff-curated threat patterns and indicators that the bot matches against live. Bait / honeypot system — the bot posts believable "help me" decoy messages to a bait channel via auto-created webhooks, then catches scam auto-responders that reply and reports them. Reporting & moderation — owners view per-guild reports, appeals, and audit trails; staff manage a global report queue, users, premium grants, and global bans. Pro features — per-channel detection aggression, trusted roles, shadow detection, alert webhooks, and CSV report export (gated by premium entitlements). Architecture Component Stack Web (web/) Laravel 12, PHP 8.2+, MariaDB, Tailwind CSS 4, Alpine.js 3.14, Vite 7 Bot (bot/) discord.py, SQLite local cache Sync HMAC-SHA256 signed internal REST API (bot ↔ web) Auth Discord OAuth (Socialite) AI Ollama (local llama3, cloud API fallback) Multi-agent detection runs in bot/detection_engine_v3.py; reports flow to the web via /api/internal/v3/*. Encryption at rest — sensitive columns (report reasons, evidence, webhook URLs) use custom Encrypted / EncryptedArray casts. Monetization — Patreon-backed subscriptions/plans. V3 design Repositioned from "Discord moderation bot" → "Scam Intelligence Platform" (branch v3, started 2026-06-16). Palette: Midnight Navy #0A192F / Electric Blue #3B82F6 / Cyber Cyan #06B6D4. Public V3 landing is now the site root (/); V2 dashboard, staff, and internal API are kept for un-ported features and bot compatibility.