Anthropic's 'Running Auto Mode in Production' Gives the Operational Playbook: 9x Longer Between Interruptions, ~10% Denial Rate at Gusto, Interactive Mode for Terraform
The companion post to the auto mode default reports that across all Claude Code usage, sessions now run 9x longer between interruptions than under the previous permission default, and that at Gusto roughly 10% of session transcripts contained at least one auto-mode denial — the classifier is firing without stalling legitimate work. The recommended pattern is layered, not classifier-only: Gusto routes MCP traffic through a governed proxy with tool guards and prompt inspection before auto mode ever sees an action, Nuro disables recursive deletes outright in settings, and Garner Health blocks Slack and email sending on the user's behalf. Teams still flip back to interactive mode for production infrastructure — Terraform, AWS, direct POSTs against live APIs — and Gusto's Evan Magnussen names telemetry as the precondition: without it, 'that would be very dangerous.'
↳ Follow the thread