When you connect an AI assistant like Claude Code or Cursor to Frontic through MCP, it can now recognize which actions are safe to run on its own. Read-only actions - looking up projects, listings, locales, storages, or previewing changes - no longer require your approval every time. Actions that create or update data still ask first, and anything that deletes data continues to require explicit confirmation.
One mail when something ships.
Product updates in your inbox, roughly once a week, never more.
Unsubscribe any time, one click.