Claude Code Auto Mode (Default on Pro, Max, and Team)

Claude Code stops asking you to approve each command and auto-approves the safe ones, for Pro, Max, and Team users.

Every tool call goes through a classifier that blocks anything irreversible, destructive, or aimed outside your environment, and Claude either finds a safer route or asks you directly
The classifier also screens for attempts to pull your data out and for hidden instructions planted in files or web pages Claude reads
Three blocks in a row, or twenty in one session, and Claude Code drops back to asking for manual approval
Becomes the default for new sessions on Pro, Max, and Team on August 14, 2026, and Anthropic says users approved 97% of the prompts it replaces
Enterprise, API, AWS, Google Cloud, and Microsoft Foundry stay opt-in for about another month, and admins can turn it off or set an org-wide default
Switch modes with Shift+Tab in the command line or the mode dropdown on desktop