I feel like you could have correctly stated this a few months ago, but the way this is "solved" is by multiple agents that babysit each other and review their output - it's unreasonably effective.
You can get extremely good results assuming your spec is actually correct (and you're willing to chew through massive quantities of tokens / wait long enough).
And unreasonably expensive unless you are Big Corp. Die startups, die. Welcome to our Cyberpunk overlords.
> You can get extremely good results assuming your spec is actually correct
Is it ever the case that the spec is entirely correct (and without underspecified parts)? I thought the reason we write code is because it's much easier to express a spec as code than it is to get a similar level of precision in prose.