Anyone else having Claude do code reviews?
I've been having Claude review my code since last month, and it's way more thorough than I expected. But sometimes it's so strict that it feels like it ends up taking more time... How are you all using it?
10 answers
Yeah, I use it too — it's really meticulous. But sometimes it points out even the smallest things, which makes it more exhausting.
This is right lol. I was using it for code review, but it just took forever because of being stuck on a specific style.
Well... I don't think AI reviews can replace human reviews. The way they grasp context is different.
I hook it up to CI and use it to catch only P0 issues. If I had to direct it one by one, the time spent giving instructions would outweigh the time spent waiting for reviews.
Got a source? Not just asking out of curiosity—I'm genuinely curious. If there's a benchmark or something, let me know.
If you put 'only the key issues' in the prompt, it's a bit better. Why would you go out of your way to dig through everything?
Ha, I'm the opposite—sometimes I deliberately write crappy code and throw it at Claude to see how much it catches, lol.
Honestly, last time they couldn't catch any serious bugs. It's just nitpicking over lint and naming conventions.
Oh, I was curious about this, so I should try it too. Though there's a saying that if you just dive in blindly, you'll end up regretting it.