AASISH-06/Agentic-AI-Phishing-Defence ? reverse-engineered prompt

Build me a Python web app that watches Gmail and alerts me about phishing in real time. It should also be able to analyze WhatsApp and SMS style messages, flag suspicious ones, and show everything in a clean dashboard with a simple risk score, sender info, and timestamps.

I want it to use the provided trained model files for phishing detection, plus extra checks like keyword spotting, OCR for text inside images, and basic steganography detection. It should store scan results in SQLite, refresh the dashboard automatically, and send alerts through WhatsApp or SMS when something looks dangerous. Also include the Gmail OAuth setup, a background worker that checks mail on a schedule, and a way to retrain the model later if needed.

Please make it easy to run locally with environment variables and example config files, and if you need to confirm anything, look up the current docs online.