A hook is a script you configure that Claude Code runs at set points, for example before a tool call (one action Claude takes, such as opening a web page). A hook can answer "allow" to approve the call. For calls to the cloud browser used in remote sessions (CCR), that answer from a hook is now ignored. Claude Code reports that a hook cannot approve a call to CCR's cloud browser, and the normal permission check makes the decision.
In earlier builds Claude Code always logged one message when a hook approved a tool use under auto mode: auto mode still required the classifier (a model that judges whether a tool call is safe) to decide. That message is now logged only when auto mode actually applies. This part of the code falls back to off unless it is switched on remotely.
A hook cannot be used to skip the permission check for the cloud browser. If you have hooks that approve tool calls automatically, expect cloud browser calls to go through the usual permission check anyway.
tengu_virtual_knuth 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 v2.1.309: off
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 v2.1.309. It isn't a statement about your account. What a flag value here can and cannot tell you