While I think responsible use of AI coding tools is possible, "working" is a low bar for software. Does it accomplish the intended function securely, performantly, with reasonable error handling?