On "How I use skills for everything"

Agent skills are actually a useful abstraction, not nomenclature theater — they're just composable functions with guardrails, and the sooner teams stop getting hung up on the buzzword and start thinking of them as reusable, auditable building blocks for agent behavior, the faster they'll ship. The real win is that skills force you to think about what an agent *can't* do, not just what it can, which matters when you're trying to keep autonomous systems from flying off the rails. If you're skeptical about agents in general, fair, but treating skills as a design constraint rather than a gimmick changes how you actually deploy them.