FraudGuard.AI

FraudGuard.AI is a scam-call simulation and debrief app. This repo contains the monorepo (frontend + backend).

Tech Stack

• Frontend: React + Vite + TypeScript
• Backend: Express + TypeScript
• Live transcript: WebSocket
• Storage: In-memory session store (no database yet)
• Twilio: Placeholder integration for voice webhooks
• Config: Environment variables via dotenv

How to Install

# From repo root
npm run install:all
# or
npm install

This installs dependencies for the root workspace and for frontend and backend.

How to Run Locally

1. Install all deps:

   npm run install:all

2. Copy environment variables:

   cp .env.example .env

3. Start both frontend and backend:

   npm run dev

   • Frontend: http://localhost:5173
   • API: http://localhost:8080

   If Twilio creds are set in .env, starting a session places a real outbound call to the phone number.

4. Optional: Use the Home page to enter a phone number, pick a scenario, and click Start. You’ll be taken to the Live page (streaming transcript), then after 10 seconds to the Debrief page (score + transcript + risk label).

How to Use ngrok for Twilio Webhooks

When you’re ready to receive real Twilio voice webhooks:

1. Install ngrok.
2. Expose your local API:

   ngrok http 8080

3. Copy the HTTPS URL (e.g. https://abc123.ngrok.io) into .env:

   PUBLIC_BASE_URL=https://abc123.ngrok.io

4. In the Twilio console, set your voice webhook URL to: https://abc123.ngrok.io/api/twilio/voice

(Actual ElevenLabs + Groq integration will plug in later; the code has comments marking where.)