A few days ago, my CTO Chris shared Cloudflare's open-sourced OAuth 2.1 library that was almost entirely written by Claude.

What caught my attention wasn't just the technical achievement, but that they'd documented their entire creative process. Every prompt, every iteration, every moment of human intervention was preserved in git commit messages—creating what felt like an archaeological record of human-AI collaboration. Reading through their development history was like watching a real-time conversation (and sometimes struggle) between human intuition and artificial intelligence.

Around the 40-commit mark, manual commits became frequent—styling, removing unused methods, the kind of housekeeping that coding models still struggle with. It's clear that AI generated >95% of the code, but human oversight was essential throughout.

Reading through these commits sparked an idea: what if we treated prompts as the actual source code? Imagine version control systems where you commit the prompts used to generate features rather than the resulting implementation. When models inevitably improve, you could connect the latest version and regenerate the entire codebase with enhanced capability.