THREAT INTELLIGENCE & API

Layer 7
Spam Scanner.

A community-driven security tool built for Filipino users to analyze URLs, SMS texts, and screenshots for phishing and smishing threats.

Stack FastAPI / React
Database Supabase
Integrations Google API
Focus Security Utilities

How It Works

Layer 7 provides users with an immediate 0–10 risk score alongside a human-readable verdict (e.g., "Verified Safe", "Be Careful", "Scam Alert"). It operates on three distinct scanning layers to ensure accuracy and adapt to emerging threats in the Philippine digital space:

01

Google API Sync

Cross-references inputs against the Google Safe Browsing database for known malicious domains.

02

Heuristic Engine

Analyzes syntax for common local smishing/phishing patterns (e.g., specific bank impersonations).

03

Community Score

Aggregates anonymous user feedback to dynamically boost risk scores on unreported, zero-day scams.

System Architecture

The project is strictly decoupled. The frontend is an optimized Vite/React application featuring the scanner UI and an open metrics dashboard. The backend is powered by Python via FastAPI, utilizing Supabase for persistent data and telemetry.

Client App

Vite + React

  • → Multi-modal UI
  • → Result Visualization
  • → Metrics Dashboard
Engine
Backend

FastAPI

POST /api/verify
POST /api/feedback
GET /api/metrics/summary
↓ Supabase DB ↓ Google API

API & Data Flow

  • 1
    Core Verification (/api/verify):

    Handles the primary ingestion of content. Includes in-memory rate limiting and basic content filtering to prevent abuse. It queries external security feeds and calculates the final 0-10 risk assessment.

  • 2
    Telemetry & Crowdsourcing (/api/feedback):

    Allows users to report results directly from the React UI as "safe", "suspicious", or "phishing". This mutates the internal community_score weight without exposing PII.

  • 3
    Public Dashboard (/api/metrics/summary):

    Exposes non-sensitive telemetry—scans per day, risk distribution, Google Safe Browsing hit rates, and total feedback counts—powering the frontend data visualization layer.