Leaving it here, because it sounded really interesting - heard there were notifications upon the agent waiting for further input and/or finishing the task in Claude Code (and maybe Codex)

It's really not a priority, but sounded cool to have :)