prishashinde52-pixel/deepfeck_recognizer — reverse-engineered prompt

Build me a simple browser app called Deepfake Recognizer that lets someone upload a photo or video and get a quick review for possible deepfake style manipulation.

It should work locally in the browser without uploading the user’s file anywhere. After they choose a file, show a clear risk score, a plain English verdict, and a breakdown of signals like texture noise, compression mismatch, lighting imbalance, and video flicker if it’s a video. Keep the UI easy to understand for non technical users, with recent report summaries so they can compare past checks.

This doesn’t need to be a real trained AI model. Make it honest that it uses forensic style heuristics and the result is only a review signal, not proof. Set it up so I can run it locally with npm start or open the HTML file, and make sure it’s ready to deploy on Netlify.