I built a way for AI to remember how you like things done

I kept running into the same issue with AI: every conversation starts from zero, even when I already solved the same problem before. I still had to repeat constraints, preferences, and workflow every time. So I built AEP (Agent Experience Protocol), a simple way to save successful AI workflows such as intent, constraints, preferences, steps, failure traps, and success checks, and reuse them later so future tasks start aligned instead of from scratch. It lives as repo-local JSON in .agent/aep/, is agent-agnostic, and works more like “Git for how you work with AI” than prompts or memory. Curious if others are hitting the same problem and whether something like this would be useful in tools like Cursor, ChatGPT, or Claude.