What this answers
Ship surfaces live workspace state two ways: the home dashboard's priorities list, and the dashboard_get MCP tool your agent calls to answer "what is in flight right now?". There is no standalone Analytics or Live-system page. The four DORA numbers tell you how the team is doing over weeks; the priorities feed and dashboard_get tell you how the team is doing right now — whether the morning ran, whether anything is waiting on a human. If you check one place before your first meeting of the day, this is the place.
The priorities feed
The middle column of the workspace home dashboard is the priorities list, served by the dashboard_priorities router. It is the closest thing Ship has to a "what needs me" panel: it pulls together what is in flight, what is stalled, and what is waiting on a person, ordered so the things that most want your attention sit at the top.
You read it the same way you would read a morning standup: scan from the top, stop at the first thing that looks wrong, click in. Most mornings nothing does, and you close the tab.
Asking your agent
The same state is available to the operator's agent through the dashboard_get MCP tool. Ask "what's in flight?" in chat and the agent calls dashboard_get (often alongside runs_list and ticket_list) and reads you back the answer — recent runs and their outcomes, what stalled, what is waiting on a human. This is the headless equivalent of glancing at the dashboard: you get the health story without opening a screen.
What you used to read here
Earlier Ship builds rendered a single "Live system" panel with a success-percentage masthead and cells for knowledge ingestion, active routines, the daily retro queue, and specialist counts. That panel and the Analytics page it lived on were removed in the MCP-edge rework. The signals it used to surface are still real and still available — through the priorities feed, the dashboard_get MCP tool, the Knowledge page for ingestion state, and the Process editor for routines and specialists — they just no longer live behind one rendered dashboard tile.
How to read it together
A healthy morning reads like this: the priorities feed is short or empty, recent runs succeeded, nothing is parked waiting on a human. A morning that wants your attention reads differently — a stalled run near the top of the feed, a clarification waiting, a routine that did not move its ticket. The story is in the priorities list or one dashboard_get call away, and it usually takes less than thirty seconds to find.
The point is not to give you another dashboard to monitor. It is to let you stop monitoring — to give you one feed you can glance at, accept, and move on from. Most days, you should.