Claude Code users will now have access to a new permissions feature called “auto mode,” which aims to enhance productivity while maintaining security. This feature, launching as a research preview for Claude Teams users, allows for longer task execution with fewer interruptions compared to previous methods that entirely bypassed permission protocols.
Under the traditional configuration, users faced strict approval requirements for each command, but “auto mode” introduces a balance by allowing safe actions to proceed automatically while blocking potentially harmful ones. A classifier evaluates each action prior to execution, ensuring that risks such as mass file deletions or data breaches are mitigated.
Enterprise and API customers can expect access to this feature in the coming days. The company continues to advise users to implement “auto mode” in isolated environments due to the inherent risks that remain even with enhanced safety measures.