It's doubly disappointing because just 4 days ago the project wrote [1], on using AI for the new Messaging app:
Why? Using an LLM as an additional reviewer seems like a good use of LLMs? Personally I have found LLMs very useful for that and they catch issues that other experienced programmers do not always find. It’s not like they are vibecoding a messages app.
What do you mean "why"? For the exact reason he stated in his first paragraph - a serious regression impacting his workflow.
... which seems to have nothing to do with using an LLM to do extra review.
My comment makes it pretty clear what the disappointment is:
> shame an obvious regression happened for one of my primary use-cases: copying one-time codes
Then the "later" is that even with frontier AI review tools, it was still missed
I am not alleging vibe coding - that's a strawman.