Post

I'm beginning to suspect that a key skill in working effectively with coding agents is developing an intuition for when you don't need to closely review every line of code they produce. This feels deeply uncomfortable!

It does, but I think it just indicates a bifurcation of the kinds of software LLM systems build vs humans; they write extremely convoluted, hyperdocumented, fairly badly designed code, but it often works. Whether that’s good or bad depends on the project, taste, workload of the future maintainers.