●AUTO — From August 14, auto mode becomes the default in Claude Code for Pro, Max, and Team. It only stops for actions judged irreversible, destructive, or aimed outside your environment●SAFETY — In a study with 1,053 paid testers, auto mode caught 89% of harmful actions while human review caught 13.6%●HABIT — Manual review becomes habitual, Anthropic notes: users approve 97% of the permission prompts Claude Code shows them●GUARD — Prompt injection screening and customizable hard deny rules have been added to keep things like data exfiltration off the table●VOICE — The head of Claude Code says he and his team have used auto mode exclusively for months and cannot imagine going back to permission prompts●VERSION — The latest release is v2.1.226 from August 8, bug fixes and reliability only. Workbench retires August 17, and Sonnet 5 promo pricing runs through August 31●AUTO — From August 14, auto mode becomes the default in Claude Code for Pro, Max, and Team. It only stops for actions judged irreversible, destructive, or aimed outside your environment●SAFETY — In a study with 1,053 paid testers, auto mode caught 89% of harmful actions while human review caught 13.6%●HABIT — Manual review becomes habitual, Anthropic notes: users approve 97% of the permission prompts Claude Code shows them●GUARD — Prompt injection screening and customizable hard deny rules have been added to keep things like data exfiltration off the table●VOICE — The head of Claude Code says he and his team have used auto mode exclusively for months and cannot imagine going back to permission prompts●VERSION — The latest release is v2.1.226 from August 8, bug fixes and reliability only. Workbench retires August 17, and Sonnet 5 promo pricing runs through August 31