Our #neathack submission Software breaks every day. The real problem is what happens after the failure. An AI can suggest a fix, but how do you know that fix actually works? And what stops the same bug from coming back? That’s what I’m exploring with Patient Zero 🧬, my solo build for #neathac under Team Trace Element. Patient Zero deliberately tests an application with unexpected inputs, investigates failures, attempts repairs in a quarantined copy, and only promotes a fix if it passes the required checks. But I wanted to take it a step further. Every verified fix can leave behind a regression test, an antibody designed to catch the same problem if it returns. The goal isn’t just to have AI write code. It’s to make fixes prove themselves before they earn their way back into the application. 🛠️ The build: • @neatlogs tracks the healing workflow, from detection to promotion. • @EntireHQ records the development process through checkpoints. • @arithecfo helped me explore the business side, including pricing, costs, and runway. I’ve deployed Patient Zero on AWS, and the live version is ready to explore. It’s not perfect. One case still slips through, and that’s part of what I want to investigate through testing and feedback. Try it: patientzero-heal.duckdns.org… 🔬 Demo video below I’m interested in what people find when they try to break it. 🛡️