GitHub Copilot Without Rules Does Not Just Produce Bad React Code. It Produces Expensive React Code.

Bad code is obvious. You see it in the review. You fix it. You move on. Expensive code is different. It looks fine. It passes the review. It ships. And then it costs you in ways that are hard to trace back to a single decision, a single prompt, a single session with GitHub Copilot. This is what happens when your AI has no rules. The cost is not in the bugs Most developers think about AI output quality in terms of bugs. Does it work? Does it break anything? That is the wrong quest...