I've never seen you like this 😥

===

Seriously, though, I ran Copilot and Claude both during this jam and it was equal parts smooth and frustrating. Some things went SWIMMINGLY (like redesigning my whole UI in Bevy, whose UI developer experience is non-existent). Others went okay; despite a lot of concern about software quality, I think the end result is better in quality than many of my other jams.

What went poorly was... well, the quirkiness of these systems. It's all new, running full speed ahead towards some unknown goal. Why does Copilot hardcode a 5 minute limit on their tasks? I thought I was clever by baking in user feedback into my agent loop...