Hi Product Hunt! 👋 I’m Mehmet. I don't have a formal engineering background, but I am currently building a massive stealth project (currently at 650,000 lines of code) using AI coding agents like Cursor and Claude. Very early on, I realized a massive problem: AI agents hallucinate. They create monoliths, generate technical debt, and falsely claim a task is 'Done' without testing. I got so tired of them breaking my codebase that I decided to write a strict set of rules they MUST obey. That evolved into Universal Agent OS. It forces agents to conduct a "Phase-0 Interview" with you before writing code, demands test evidence, and enforces a strict "Zero-Leak Protocol". Today, I'm launching the official VS Code Extension. I built this out of pure necessity to survive my own 650k-line project. I'd love for you to try it with your AI agents and let me know what you think!

Universal Agent OS

A strict constitution to stop AI agents from hallucinating

A strict constitution to stop AI agents from hallucinating