I’ve had several occasions where I would build a feature “properly” with defensive guardrails and handling of edge cases “that will never happen (but are technically possible)” and then you run into the coworker that says “it’s overengineered. YAGNI. Just do the one-line fix”
But...were they wrong? There is a difference between "edge case" and "cannot ever happen". I've seen this trend a lot lately where Claude suggests a lot of extra code to handle things that cannot possibly happen. Unless you ask it to, it won't trace the flow of data through the app to confirm the edge case actually exists, if local conditions appear to allow it. Then the developer lets it implement this crap without asking that one question, and I have to say YAGNI in the review.