oma Console Agent Monitor tab

Console AgentDetail shows a derived health strip (status, last/next run, uptime, success rate, avg duration, cost/run) from existing schedules, sessions, stats, and analytics. No dedicated /v1/agents/:id/health route yet.

Monitor (/agents/:id/monitor) sits between Schedules and Observability. It is the live agent.status progress feed: current or last session, step bar, amber blocked_on, heartbeat log, lag at 2× the 5-minute default. Eyes-only — Interrupt is out of this slice. Upgrade log is an honest empty placeholder.

Dashboard Active sessions deep-links to that agent's Monitor when exactly one running session has an agent id.

Landed in #444 as the first slice of the monitoring UX (#345 still open). Signed-out /monitor stays behind the Console login gate.

Why: Operators of scheduled or long-running agents need a glanceable alive/progress surface without new control-plane endpoints.

How to apply: Point operators and recert at AgentDetail header + Monitor tab, not Observability. Do not invent heartbeat_missed / self_upgrade notify until a dispatcher exists.

Hub: project-open-managed-agents.