Auto mode became Claude Code's default permission mode on August 14, and its classifier calls stopped counting toward usage limits
As of August 14, 2026, new Claude Code sessions on Pro, Max, and Team plans start in auto mode, where a background classifier approves safe actions and blocks risky ones instead of prompting; a default you set yourself persists unless you accept a one-time switch prompt, and org-managed defaults are untouched. Anthropic also confirmed the classifier calls auto mode makes no longer count against usage limits, which removes the main cost argument against leaving it on. The actionable move for anyone scripting or running unattended agents: set `permissions.defaultMode: "auto"` in `~/.claude/settings.json` rather than reaching for `--dangerously-skip-permissions`, since auto mode has hard-deny rules that the skip flag does not.
↳ Follow the thread