Interactive menus are no longer blocked in agent background sessions
You can once again use interactive menus, like changing AI models, without them getting stuck while running agent tasks behind the scenes. Previously, an overly strict guard (a protective mechanism meant to prevent unwanted actions) accidentally blocked the /model command and other dialogs from appearing in claude agents (a feature where Claude opens multiple agent instances to work for you) background sessions (tasks running silently in the background). This meant you couldn't interact with these interfaces when you needed to. We've rolled back this overly broad restriction so everything works normally again. It's like an office security door that was set so strictly it locked regular employees out of their own meeting rooms; the rule has been reverted to its standard setting so you can actually push the door open and get to work.