Anthropic has introduced a new “auto mode” in Claude Code to improve how developers handle permissions during coding tasks. The feature aims to reduce constant approval requests while still keeping systems secure. It offers a balanced approach between strict permission controls and fully skipping them.
Anthropic explains that Claude Code’s default setup requires user approval for each file write and command execution. This setup protects systems but slows down workflows, especially during longer tasks.
To avoid interruptions, some users previously chose to skip permissions entirely. However, that approach increased risk and exposed systems to unsafe actions.
Anthropic said, “Auto mode is a middle path that lets you run longer tasks with fewer interruptions while introducing less risk than skipping all permissions.”
The new mode allows Claude Code to continue tasks with fewer prompts while still monitoring actions in the background.
Before any action runs, a classifier checks whether it is safe. It looks for risky behavior such as deleting large amounts of data, accessing sensitive information, or executing harmful code.
Safe actions proceed automatically without user input. Risky actions get blocked, and Claude tries a different method. If the system repeatedly attempts restricted actions, it eventually asks the user for permission.
Anthropic said, “Actions that the classifier deems as safe proceed automatically, and risky ones get blocked.”
The company also noted, “Auto mode reduces risk compared to dangerously skip permissions but does not eliminate it entirely.”
Because of this, Anthropic still recommends using the feature in controlled or isolated environments.
Anthropic confirmed that Claude Teams users can access auto mode as a research preview starting today. Enterprise and API users will receive access in the coming days.
The update follows another recent feature where Claude can control a Mac system, also released as a research preview.
