Hooks are actions you configure to run automatically at set points. Some run when new input arrives. Claude Code now keeps reading incoming messages while those hooks run, and sets them aside to handle afterwards.
If a permission request (a question asking whether Claude may take an action) is waiting when you press Stop or the session ends, it is now withdrawn. The model is told the request was withdrawn unanswered and should not ask again. When this behaviour is switched off, the hook simply runs as before.
Stop and permission handling stays responsive during a slow hook, mainly in headless sessions (run with no interactive screen) and remote sessions.
tengu_breezy_haven Not enough to sayNothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.
This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.295: on
These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.
Read once, for one account on one subscription tier, against v2.1.295. It isn't a statement about your account. What a flag value here can and cannot tell you
New in this build: tengu_breezy_haven