Launched March 24, Claude Code's Auto Mode eliminates the constant permission prompts that frustrated developers. Previously, Claude Code would pause before every file write, terminal command, or web search — meaning you'd walk away only to return and find it waiting for approval.
Now Claude decides which actions are safe to execute autonomously. It uses AI safeguards to evaluate each action before execution and blocks prompt injection attempts. The feature currently works with Sonnet 4.6 and Opus 4.6 models only.
Anthropic recommends using Auto Mode in isolated environments as a precaution, though the built-in safety checks are designed to catch dangerous operations before they execute.
The change reflects a broader industry shift: AI coding tools are moving from suggestion engines to autonomous agents. The key challenge is trust — developers need confidence that the AI won't accidentally delete files, expose secrets, or run destructive commands. Auto Mode's approach of AI-evaluated safety checks is Anthropic's answer to that problem.
For developers who spend hours with Claude Code, Auto Mode is a significant quality-of-life improvement. No more babysitting the terminal.