Unclear What auto mode does differently for you once its classifier counts as switched off is not stated.
In auto mode, Claude Code uses a classifier, a check that decides whether an action is safe to run without asking you. Claude Code now counts that classifier as switched off when CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS is set and Claude Code talks straight to Anthropic's own API rather than going through another provider.
Before, whether the classifier's extra requests were retired depended only on a list sent from Anthropic's servers. There is now also a server-side setting with these values:
It can retire the classifier's extra requests, and it is off unless it is switched on remotely.
Setting CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS now changes how auto mode checks actions, not only which headers are sent. If you set it on a direct connection, expect auto mode's safety check to count as switched off.
tengu_drifting_valley 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 v2.1.314: 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 v2.1.314. It isn't a statement about your account. What a flag value here can and cannot tell you
What auto mode does differently for you once its classifier counts as switched off is not stated.
New in this build: tengu_drifting_valley