i'm getting a bit addicted to using #AI coding agents. the machines are pretty dumb and lie a lot about having solved problems when they actually haven't, but with persistence and a bit of manual reading and adjustment i can get to a full working solution way faster than i ever have been able to before.

my current fixation is about creating a fully comprehensive profiling report for #orly relay - i want to make it as fast as possible, and a somewhat orthogonal effect of this is that it's finding more logic bugs as i go through because these often manifest as cpu, memory or allocation blowing out and causing performance problems and at the same time, diminishing the effectiveness of the functions, which can basically be called bugs.