code-execution

Anthropic has released auto mode for Claude Code in research preview, enabling the AI to execute actions it deems safe without waiting for user approval. The feature uses built-in safeguards to block risky actions and prompt injection attacks, while automatically proceeding with safe operations.