Allows users to manage multiple remote AI coding agents from a single Claude Code session, with a controlled execution model where operations require moderator approval.
