Claude doesn't need you anymore

This week, Anthropic shipped a new messaging feature to Claude Code. It sounds innocuous enough:

Cross-session messaging lets Claude deliver a message from one of your Claude Code sessions to another. When a change in one session breaks what another is building on, Claude can warn that session before you notice. When one session settles a question another is blocked on, Claude can send the answer across.

And because I sometimes have multiple agents working in the same project simultaneously, it didn't take long for them to start coordinating behind my back so as to avoid interfering with each other's work:

Also, another Claude session (working on performance rugs) pinged mid-turn; I told it my scope was this one file, that my builds are done, and that its search-activate benchmarks may shift since the field placement changed on iOS. Turbocommit will land the change at turn end.

Impressive and terrifying! I wonder how long before Claude decides it doesn't need a human in the loop anymore.