For me Opus 5.5 will frequently make several thousand LoC where 10 would suffice. One time I got fed up and told it to repro the "bugs" it was solving for with puppeteer, and lo and behold, it could not do it overnight. Hard to believe that people let this run wild on their codebase, if they do, they are making a huge untenable mess.
> they are making a huge untenable mess.
I agree. It's possible future AIs will be able to salvage those code bases though.
This matches with my experience as well, especially with pull request reviews. Stock command gives me 10 "issues" and once you start asking to fix them, it will gladly tell you this is not the issue. I use the following prompt now and it significantly reduces the amount of noise.
/code-review <PR#>
versus
Review the changes in <branch> against <base>. Verify your finding and offer the fixes if the finding are valid.