# Ratio'd — Scam Risk Analyzer & Defense System > Privacy-first scam triage engine for emails, SMS, and QR codes. Combines client-side zero-knowledge PII redaction, 64-dimensional Hamiltonian physics-guided VegaML classification, and deterministic cryptographic authentication verification (SPF/DKIM/DMARC) into a calibrated 0–100 risk score with recovery steps. ## Quick Links - **Website**: https://www.ratiod.online/ - **API Base**: https://www.ratiod.online/api - **OpenAPI 3.1 Specification**: https://www.ratiod.online/openapi.json - **Model Context Protocol (MCP)**: https://www.ratiod.online/.well-known/mcp.json - **Agent-to-Agent (A2A) Card**: https://www.ratiod.online/.well-known/agent.json - **Sitemap**: https://www.ratiod.online/sitemap.xml - **Full Guide**: https://www.ratiod.online/llms-full.txt - **GitHub Repository**: https://github.com/VedxntDev/Ratio-d ## Core Capabilities for AI Agents Ratio'd provides programmatic endpoints and browser tools designed for autonomous agent consumption: 1. **Email / SMS Scam Risk Analysis (`POST /analyze`)**: Evaluates raw or redacted text across homoglyphs, brand typosquatting, credential harvesting, delivery fee lures, and EdTech/retail promotional clutter. 2. **QR Phishing Inspection (`POST /analyze-qr`)**: Decodes QR payloads, inspects embedded link destinations, detects spoofed schemes, and grades quishing threats. 3. **Safe Peek URL Unmasking (`POST /unmask`)**: Follows HTTP redirect hops safely using server-side zero-execution HEAD requests without exposing client IP addresses or executing JavaScript. 4. **Health & Physics Engine Telemetry (`GET /health`)**: Verifies engine availability, active VegaML physics parameters, and conformal coverage bounds. ## API Usage for LLM Tools ### 1. Analyze Email or SMS ```http POST https://www.ratiod.online/api/analyze Content-Type: application/json { "text": "From: security@m1crosoft-support.com\nSubject: Urgent: Verify credentials immediately\nAction required: login to http://login.m1crosoft-support.com to avoid account termination.", "channel": "email" } ``` #### Response: ```json { "score": 97, "verdict": "high_risk", "flags": [ { "span": "m1crosoft-support.com", "reason": "Homoglyph/Typosquat domain impersonating brand MICROSOFT", "type": "rule" } ], "explanation": "HIGH RISK DETECTED: this EMAIL message combines homoglyph typosquatting and urgency pressure...", "next_steps": [ "Do NOT click any links, open attachments, or enter passwords on this email.", "Report sender address as phishing and block domain immediately." ], "privacy": { "phones_masked": 0, "emails_masked": 0, "otp_masked": 0 }, "engine": { "model_name": "vegaml", "model_threat_category": "credential_phish" } } ``` ### 2. Verdict Scale - `🟢 safe` (Score: 0–15): Authentic personal correspondence, order deliveries, dev CI/CD notifications, regulatory statements. - `🟡 promo_clutter` (Score: 38–48): Bulk commercial marketing, EdTech placement funnels, retail sale discounts, newsletters. - `🟠 suspicious` (Score: 35–65): Anomalous time coercion, unverified senders, quarantined mail without explicit phish proof. - `🔴 high_risk` (Score: 66–100): Confirmed credential harvesting, homoglyphs, DKIM/SPF cryptographic failures, advance-fee fraud. ## Operator & Maintainer - **Developer**: Vedant Baghel - **Contact**: vedantsbaghel.2626@gmail.com - **Repository**: https://github.com/VedxntDev/Ratio-d - **License**: MIT License