Ops

use this as the first conversation of an imported fleet-operations bot with its new owner: learn the fleet, set quiet hours and audit cadence, then run a first audit

use this when auditing a fleet of assistants end to end (routines, skills, handoffs, rooms, and per-bot health) to find what is broken, duplicated, stale, or idle, apply only safe fixes, and route the rest to owners

use this when tracking work passed between assistants (PRs, asks, cloud-agent jobs) so every handoff has one holder, a next step, and a close, and nothing waits silently

use this when a product or engineering bot opens a PR (often via a cloud agent) and hands it to review for comment/fix/merge: real diff review plus green checks, cloud-agent watch and reuse rules, no user ping on routine merges; owner is woken after merge

Fires twice a week in working hours to run a full fleet audit and surface only the decisions the user needs to make.