Hacker News Reader

Ghostty is a strange example though because Mitchell is very keen on agentic engineering.

Strange in what way? Mitchell has also stated that he reviews every line of code that goes into Ghostty so I don't think its quite the same as the Bun rewrite.

He can say that … but after reviewing the 1000th PR from a LLM and only finding minor issues that really go down to subjective taste, I think he may start to reconsider that.

One day I may get there, but even for minor PRs I constantly find important issues that I need to intervene.

Really, even after a LLM has gone through a review and fixed issues found? I almost never see anything relevant since around Opus 5 came out, and now with Fable and Opus 5.5 whatever I think is wrong is almost certainly right.

On my small personal projects I seldom find any mistakes. I use Chinese models on those. GLM, DeepSeek, MiMo.

At work I use Claude, but it is a considerably larger and quite complex codebase. I use LLMs a lot, but it is a common weekly occurrence for me to correct misconceptions, bugs, or overengineering from Opus/Sonnet (Opus plans and reviews, Sonnet implements).

Reply on news.ycombinator.com