[rappatic]

Counterintuitively, I feel like this will not be super useful, at least for me. My bottleneck is MY ability to parse and understand LLM-generated code. The agent can code a lot faster than I can read and understand its output.

[altmanaltman]

People who vibe code don't really care about silly things such as understanding the code though.

[rovr138]

There's a lot of people out there that aren't paying as close attention to the actual code. Wild times!

[nonethewiser]

Smart if you can get away with it. For non trivial things you can't quite get away with it.

I've started asking Claude to make me a high-level implementation plan and basically prompt ME to write it. For the most part I walk through it and ask Claude to just do it. But then 10% of the time there is a pretty major issue that I investigate, weigh pros/cons, and then decide to change course on.

Those things accumulate. Maybe 5-10 things over the course of an MVP that I wouldn't really have a clue about if I let Claude just dutiful implement it's own plan.

[rovr138]

Oh, I don’t like it.

Not including broken things, you end up with random errors or things that just don’t work right.

[ollysb]

If it was fast I'd ask questions more than read the code in detail. This isn't viable for that approach yet though.

[airspresso]

Spend time building a test harness and evaluations of whether the solution meets the requirements. Then you don't need to look at the code because those other pieces will bring the necessary guarantees and trust.

[post-it]

I've never actually seen a test plan thorough enough to allow that kind of behaviour, and I doubt one exists.