Tag
Matt Pocock proposes documenting agreed test seams in apps, arguing that AI agents cannot be trusted to make good testing decisions, often leading to fragile tests that break on implementation changes.