I Read All of Cloudflare's Claude-Generated Commits

https://news.ycombinator.com/rss Hits: 12
Summary

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.The lead engineer, @kentonv, started as an AI skeptic. "I was trying to validate my skepticism. I ended up proving myself wrong." Two months later, Claude had generated nearly all of the code in what became a production-ready authentication library.Kenton included the prompt used to generate code in every commit, which made this exploration possible. As we begin to lean more heavily on AI tools within development, this practice will become increasingly important. Sometimes, the original prompt is more valuable (and easier) to review than the resulting code—especially when an engineer declares an incorrect assumption that the model blindly follows.This transparency transforms git history from a record of changes into a record of intent, creating a new form of documentation that bridges human reasoning and machine implementation. Reading through roughly 50 commits revealed some interesting patterns about how this collaboration actually unfolded:Prompt by example. Their initial prompt was a substantial code block showing exactly how the library would be used by a worker implementing it. This approach eliminated ambiguity about method signatures while revealing practical considerations that abstract specifications often miss. It's the difference between describing a dance and demonstrating the choreography."You did X, but we should do Y. pls fix." The most effective prompts followed a consisten...

First seen: 2025-06-07 02:09

Last seen: 2025-06-07 13:11