Anthropic is adding an auto mode to Claude Code, its AI-powered coding assistant. Currently in research preview for Claude Team users, auto mode acts as a middle ground between the default conservative permission model (which requires approval for every file write and Bash command) and the risky --dangerously-skip-permissions flag. A classifier reviews each tool call before execution, blocking potentially destructive actions like mass file deletion, sensitive data exfiltration, or malicious code execution, while allowing safe ones to proceed. The feature is set to roll out to enterprise and API users in the coming days.
Sort: