Watch
0
0
Fork
You've already forked hyperhive
0

swarm ui: show agent freshness and failures where the operator looks, page titles, try again, and stale-data notes

An agent card shows its freshness (stale with the last report's age, never reported, not in swarm identity) when it is not fresh, so the status filter has a visible reason. A failed state change shows its reason in a panel under the card, and a failed cancel shows its reason in the badge, instead of a hover title. Pages can set their title (usePageTitle) and the flow editor does, so several editor tabs differ; the 404 links home. The claude login dialog offers try again after a failure, announces its wait, and has a per-agent field id. The create-agent tabs go forward again as far as the operator has been, but not past a step that is invalid now (lastReachableStep, tested), and the disabled next and create buttons say why in a hint they are described by. usePolled reports when the last live answer arrived, and StaleNote says how old a list is under a failed refresh, or that it is the answer saved from the last visit, on the alerts, builds, review runs, hives, schedules and flows lists.
This commit is contained in:
iris 2026-11-19 18:59:20 +01:00 • committed by mara
commit 679d2ee420

View file

@ -43,7 +43,10 @@ Everything it shows comes from [`swarm-controller`](../../../swarm-controller/RE
Pages that refresh on a timer share one polling rule: the page sends no new
request while the last one is still running, so an older answer never overtakes
a slow one, and a failed refresh keeps the rows already on screen under an
error. While a page is fetching, its refresh picker shows a spinner in place of
error. A list that isn't live says so: under a failed refresh it reads "showing the
last answer, from" and the time, and a page opened on the answer saved from the last
visit says that until a fresh one arrives (the alerts, builds, review runs, hives,
schedules and flows lists). While a page is fetching, its refresh picker shows a spinner in place of
the clock, for at least 0.4 seconds so a quick answer still reads as a refresh.
On a folded card (jobs, builds, hives) the picker sits in the card's header, left of the
chevron, and shows once the card is open; selecting it doesn't fold the card.
@ -72,7 +75,8 @@ An agent created here or with `swarmctl agent create` starts `paused`; set it
`up` from its card. A card shows **login needed** when the agent's own
turn-state header says its claude has no session, and a **log in** button
that runs the claude login from a dialog: open the URL it shows, then paste
the code back. When the secret store has the login, the dialog closes and a
the code back. A failed login offers **try again** in the same dialog. When the
secret store has the login, the dialog closes and a
short toast in the top-right corner says so. The credentials go to the agent through the secret store
([`claude-login.md`](../../2-handbook/swarm/claude-login.md)). The agent's own web UI login still
works too. The settings section of an opened card has a **claude login** row: when
@ -80,6 +84,11 @@ the swarm stored the agent's login, or that none is, with a **log in** button
(**log in again** when the swarm holds one) that opens the same dialog, and a **delete
login** action that asks first.
A card shows its freshness when the agent isn't **fresh**: **stale** with how long ago the hive last
reported it, **never reported**, or **not in swarm identity**. A fresh agent shows nothing, so the
badge is the reason a card matched the status filter. A failed state change shows its reason in a panel
under the card's header, not in a hover title.
A card shows **updating** when the agent's status snapshot has `updating` set.
The hive sets it when an update stops a running container and clears it when
the tail reconcile or start settles the agent, whether the container came back
@ -537,7 +546,7 @@ error and the agent gets nothing.
While the agent is in a turn, a **cancel turn** badge appears among the state
badges over the terminal, in its own tab. It asks the agent's harness to stop
the turn, then shows whether one was running ("turn cancelled" or "nothing was
running"); a failure shows "cancel failed" with the reason on hover, and a click
running"); a failure shows "cancel failed: " and the reason in the badge, and a click
tries again. Subagents have no message box and no cancel badge.
The message doesn't go through the agent's hive, so the hive dashboard's