Я провел 42 дополнительных спринта агентов на более сложной задаче. Вот что не исправил рабочий процесс.

Оригинал

I ran 42 more agent sprints on a harder task. Here's what the workflow didn't fix.

In the first post a Scrum workflow for coding agents looked clean: with the skill, 6 of 6 runs met every item of the Definition of Done, against 3 of 6 without it. Several of you pointed out the obvious catch. One small task, one model, three runs per side. That's a hint, not a result. So I ran a bigger one, and I wrote down the rules before I looked at a single run. What changed A harder sprint. A new Go repo with four items that depend on each other across modules. Two of them c