I’m trying Claude Code (CC) after reading [AI Changes Everything][1] by @mitsuhiko and my initial reaction is that I’m a bit baffled.
It keeps suggesting small, individual changes (diffs) to specific lines of specific files, and asking me to approve them or not. Most of them are OK, some of them are not. Whatever. But I have to sit here twiddling my thumbs between each interaction. It’s tedious. I could do this work faster on my own, and I’d have higher confidence that it’d actually work. (Some of the changes will need verification — especially because I’m running CC on a “quick and dirty” “temporary” project that has no automated tests.)
I’m guessing maybe Ronacher has CC configured to not prompt about each individual change, but to just go ahead and work through an entire change — in other words, the review/update/review cycles are larger, coarser-grained. So maybe I’ll try that.
