Anthropic rolls out Claude Code auto mode with safety checks
Anthropic just rolled out "auto mode" for its Claude Code AI tool, making it easier for developers by letting the AI decide on permissions with built-in safety checks.
It's a safer upgrade over the "dangerously skip permissions" setting and is currently in research preview for Team plan users, with a wider release on the way.
Anthropic's Claude Code inspects tool calls
Auto Mode uses a classifier to review each tool call to check for potentially destructive actions like mass deleting files or sensitive data exfiltration while running safe ones automatically.
Risky ones are blocked, prompting Claude to take a different approach, and repeated blocks eventually trigger a permission prompt to the user.
There might be a small impact on speed or token use, but overall, this feature aims to make AI coding tools both efficient and secure, something that could really shape how we code in the future.