Claude Code Monitor caps at 30 min: pair it with a background-Bash waiter for multi-hour jobs

The Monitor tool expires after at most 30 minutes, so a per-event log watch on a multi-hour remote job must be re-armed on each expiry, and events between expiry and re-arm are lost if you use tail -n0. For the one event that must not be missed (job finished), use a separate run_in_background Bash loop that polls a status file over ssh every few minutes and exits on the terminal state; it has no 30-minute cap. Keep the Monitor only for the per-attempt progress pings.