docs/web-ui/dashboard.md

@@ -209,12 +209,8 @@ Three sub-tabs: **R3BU1LD QU3U3** (default), **M3T4 1NPUTS**,
 **BUILD L0GS**. Its own esbuild bundle (`builds.js`); cold-loads
 `/api/state` and subscribes to `/api/dashboard/stream` for
 `rebuild_queue_changed`, `meta_inputs_changed`, `meta_update_running`.
-The SW4RM tab (`/dashboard.html`) does **not** read this endpoint (or
+The dashboard tab keeps the rebuild-queue *state* for the SW4RM
-any job-queue state) at all — mara, on review: "swarm.js should not
+card badges without rendering these panels.
-need to pull in the jobq to do its job." Its per-agent pending badges
-are transient-only, and its queue-summary banner is removed for now
-(see Container row, below), pending a dedicated rollup endpoint it
-can read directly instead of the full graph.
 
 **R3BU1LD QU3U3** — pending, in-flight, and recently-settled container
 operations: rebuilds, meta-update cascades, and first-spawns. One
@@ -226,11 +222,9 @@ listens for its `hive-jobq-graph-update` event to drive the two things
 below it that the generic view doesn't show. The component owns
 fetching, cold and live: `GET /api/jobq/graph` on mount, and
 `.refresh()` on every `rebuild_queue_changed` SSE tick (that event
-still carries its own `Vec<QueueEntry>` payload — `DagView`-shaped,
+still carries its own `Vec<QueueEntry>` payload on the wire — the
-also read by `hivectl`'s own wait/progress loop, a separate migration
+SW4RM tab still consumes it for the badges, untouched — this page just
-— on the wire, but neither dashboard page reads it anymore; both
+ignores the payload and treats the tick as a refetch trigger).
-treat the tick as a pure refetch trigger against the generic
-endpoint).
 
 Each row is one root graph node (`parent: null`); a multi-step op's
 per-agent subgraphs and sub-steps render as nodes within that one
@@ -885,19 +879,22 @@ fetch entirely.
   independent badge rather than being collapsed into one label,
   matching the existing multi-badge convention this line already uses
   for `paused`/`needs_update`/model/ctx.
-  Any pending badge means the row is actually **running** something
+  The row visual splits queued vs running: a **queued** entry (no
-  right now — there is no separate queued-but-not-started row state
+  transient yet, see below) shows only the pending-state pill (no row
-  to visually distinguish it from (see **Pending-badge derivation**
+  tint, so a long queue doesn't paint half the tab amber); a
-  below), so every row carrying ≥1 badge keeps the amber row tint AND
+  **running** entry (at least one transient) keeps the amber row tint
-  draws a **rotating amber ring** around the agent icon.
+  AND draws a **rotating amber ring** around the agent icon, so it's
+  obvious at a glance which container is actually moving.
 
-  **Pending-badge derivation:** transients only (`transientsState`,
+  **Pending-badge derivation:** two separate stores, but no longer a
-  keyed `agent -> Map<kind, since_unix>`) — a transient is **derived
+  priority *order* between them — the second only ever applies when
-  from a job-queue node currently `Running`** against that agent, not
+  the first has nothing to say. (1) **Transients**
-  declared per request, so its label follows the operation as it
+  (`transientsState`, keyed `agent -> Map<kind, since_unix>`) — a
-  progresses (a rebuild reads `stop_for_update`, then `swap`, then
+  transient is **derived from a job-queue node currently `Running`**
-  `reconcile` rather than one constant `rebuilding` for its whole
+  against that agent, not declared per request, so its label follows
-  life). Two consequences for anything rendering it:
+  the operation as it progresses (a rebuild reads `stop_for_update`,
+  then `swap`, then `reconcile` rather than one constant `rebuilding`
+  for its whole life). Two consequences for anything rendering it:
 
   - The label vocabulary is **open** — it is the node's own wire tag
     (`NodeKind::as_str`, the same strings `NodeView.kind` carries),
@@ -915,16 +912,20 @@ fetch entirely.
   their own label directly via `TransientSet`/`TransientCleared`
   events carrying no backing node at all.
 
-  **Queued (not-yet-started) work shows nothing on the card, and
+  (2) The **rebuild-queue fallback** (`rebuildQueueState`) only
-  `swarm.js` reads no job-queue state at all.** An earlier version of
+  fires when an agent has **zero** transients — since (1) now covers
-  this page had a job-queue-backed fallback badge for the `Pending`
+  every `Running` node unconditionally, a `Running` rebuild-queue
-  case (`queuedOpsByAgent()`, reading a `GET /api/jobq/graph` fetch);
+  entry can never usefully reach this fallback by the time it's
-  removed per mara, on review of hyperhive#3028's PR: *"swarm.js
+  consulted; the fallback exists purely for the **`Pending`
-  should not need to pull in the jobq to do its job,"* followed by
+  (queued, not yet started)** case, which `running_transients()`'s
-  *"remove the per agent pending stuff - only show what is running."*
+  Running-only test cannot represent. `queuedOpsByAgent()` (swarm.js)
-  The queue-summary banner below was, for the same reason, also pulled
+  reflects this: it only ever looks at `Pending`-state queue nodes,
-  rather than kept on that same client-side graph-fetch — see its own
+  and (mara, on review) the badge shows the queue entry's raw `kind`
-  entry below for why.
+  string as-is — no English-phrase lookup translating it first, same
+  opaque-string treatment a transient's own label already gets.
+  `opRunning` (driving the `pending-running` row class + spinner) is
+  simply "does this agent have at least one transient" — a queued-only
+  entry (no transient yet) leaves it false.
   An **active model badge** (`model · <name>`, blue) appears when the
   container is running and the harness has persisted a model name
   (`harness/hyperhive-model`). Read by hive-c0re's `ContainerView`
@@ -965,23 +966,13 @@ per-agent actions and navigation links. Contents:
 agent is stale. Banner pulses on each broker SSE event
 (`pulseBanner` with a 4s grace timer).
 
-**Build-queue summary banner — currently removed, coming back on a
+**Build-queue summary banner** — when the rebuild queue has any
-different data source.** Previously a compact amber banner above the
+`queued` / `running` entries, a compact amber banner sits above the
-container list (`◐ build queue — N running · M queued — view
+container list: `◐ build queue — N running · M queued — view queue →`
-queue →`, linking to the BU1LDS page), derived client-side from `GET
+(the link goes to the BU1LDS page's R3BU1LD QU3U3). It replaces the
-/api/jobq/graph` (filter to `parent: null` group roots — each an
+old per-transient spinner list; the actual running node for each
-independent operation, not a raw node count — bucket by `state`).
+agent is already shown on its card (transient + in-flight-queue
-Pulled per mara, on review of hyperhive#3028's PR: *"dont replace one
+badges), so the top of the tab only needs the at-a-glance summary.
-legacy thing with another. then we will have to either wait with
-this pr or split it into what can and cannot be done now"* — a
-client-side derivation over the generic graph was itself judged a
-stopgap not worth landing, the same way the old `DagView` read it
-replaced was. Comes back once a dedicated rollup endpoint (status ×
-count pairs, tracked separately) exists for `swarm.js` to read
-directly instead of pulling the whole graph in for a summary. Until
-then, the actual running work per agent is still visible on each
-card via the transient pills; only the hive-wide at-a-glance count is
-missing.
 
 ### Themed dialogs
 

frontend/packages/dashboard/src/dashboard.css

@@ -347,11 +347,8 @@ hive-agent-menu {
 .container-row:hover hive-agent-menu {
   --menu-btn-opacity: 1;
 }
-/* Card actions dim while a row has any pending-state badge (transient-
+/* Pending state splits queued vs running. */
-   driven only now — see swarm.js's pending-badge derivation comment;
+.container-row.pending .actions { opacity: 0.4; pointer-events: none; }
-   there is no separate queued-but-not-running row state to tell apart
-   from this one anymore). */
-.container-row.pending-running .actions { opacity: 0.4; pointer-events: none; }
 .container-row.pending-running {
   border-color: var(--amber);
   background: color-mix(in srgb, var(--amber) 5%, transparent);
@@ -458,11 +455,7 @@ hive-agent-menu {
 
 /* Build-queue summary banner on the SW4RM tab: one compact line
    when the rebuild queue has active work, with a link to the full queue on
-   the C0R3 page. Amber to match the in-progress / "rebuilding" card tint.
+   the C0R3 page. Amber to match the in-progress / "rebuilding" card tint. */
-   Currently unused — swarm.js dropped the banner pending a dedicated
-   rollup endpoint (see swarm.js's transients-section comment) — kept
-   here rather than deleted-then-restored, since the markup/classes are
-   expected to come back unchanged once that endpoint exists. */
 .queue-summary {
   display: flex;
   align-items: center;

frontend/packages/dashboard/src/swarm.js

@@ -1,7 +1,7 @@
 // SW4RM (containers) domain — extracted from tabs.js.
 // Agent topology, container-row rendering, selection bar, peer-hives block,
-// and all live-update apply handlers for container-state and transient ops.
+// and all live-update apply handlers for container-state, rebuild-queue, and
-// See docs/web-ui.md::Container row for the rendering contract.
+// transient ops. See docs/web-ui.md::Container row for the rendering contract.
 
 import {
   $, form, fmtAgeSecs,
@@ -27,6 +27,8 @@ const CTX_CAUTION_TOKENS = 100_000; // fallback yellow threshold (~= 50% of 200k
 
 // ─── module-level state ─────────────────────────────────────────────────────
 
+let rebuildQueueState = [];
+
 // Keyed container row cache. Maps agent name -> { el: <li>, fingerprint }.
 // Allows renderContainers to skip rebuilding rows whose displayed state
 // hasn't changed — prevents full-wipe flicker + avoids redundant async
@@ -50,20 +52,62 @@ const transientsState = new Map();
 // tab-gated visibility).
 const selectionState = new Set();
 
-// ─── transients ─────────────────────────────────────────────────────────────
+// ─── rebuild queue ──────────────────────────────────────────────────────────
+
+export function syncRebuildQueueFromSnapshot(s) {
+  rebuildQueueState = (s.rebuild_queue || []).slice();
+}
+export function applyRebuildQueueChanged(ev) {
+  rebuildQueueState = (ev.queue || []).slice();
+  // Re-render the SW4RM tab so newly-queued ops light up the right
+  // card with a "<kind> queued…" badge, and entries that drop out of
+  // Pending fall back to whatever transientsState (or nothing) says
+  // instead. Running work is *not* driven by this event — that's
+  // transient_set/transient_cleared's job, since every running node
+  // naming an agent already lights a pill by the time it gets here.
+  // See docs/web-ui.md::Container row for the badge taxonomy.
+  renderContainersFromState();
+}
+// Map from agent name -> the queued op's raw `kind` string, backing
+// the SW4RM card's fallback pending badge. Pending only — every
+// *running* node naming an agent already lights a transient pill (any
+// running node, not just a curated "worth it" subset — see
+// docs/web-ui.md::Container row), so a Running entry here would
+// always be redundant with `transientsState` by the time this is
+// consulted. Queued (not yet started) work is the one state
+// transients can't represent, since `running_transients()` on the
+// backend is a Running-only test.
 //
-// This is now the ONLY job-queue-derived state on this page — mara, on
+// The returned `kind` is displayed as-is (mara, on review: "drop
-// review of the DagView migration: "swarm.js should not need to pull in
+// queuedLabelFor - just show what the backend sends") — same opaque-
-// the jobq to do its job." An earlier version of this file also fetched
+// string treatment `pending`'s transient half already gets, no
-// GET /api/jobq/graph directly for two things now gone: a per-agent
+// English-phrase lookup table translating it first.
-// queued-but-not-running badge (`queuedOpsByAgent()`, removed per "remove
+//
-// the per agent pending stuff - only show what is running" — a card's
+// Agent is per-node, not per-DAG (a DAG can span agents — e.g. the
-// pending badges are transients-only now, which already means "what is
+// startup sweep's MetaLock cascade, or a hive-wide restart), so this
-// running") and the queue-summary banner (removed per "dont replace one
+// derives each agent's queued state from its own node(s) within the
-// legacy thing with another" — a client-side derivation over the generic
+// entry rather than the DAG's overall `state`/`kind`.
-// graph was itself judged a stopgap not worth shipping; the banner comes
+function queuedOpsByAgent() {
-// back once the dedicated rollup endpoint exists, and reads that
+  const out = new Map();
-// directly instead of pulling the graph in here at all).
+  for (const e of rebuildQueueState) {
+    if (e.state !== 'Pending') continue;
+    // spawn ops target an agent that doesn't exist yet as a
+    // container — the transient store already drives the
+    // pending row for that case. Skip here to avoid double-
+    // surfacing if the spawn op happens to land in the queue
+    // while the row exists transiently.
+    if (e.kind === 'spawn') continue;
+    for (const n of e.nodes || []) {
+      if (!n.agent || n.state !== 'Pending') continue;
+      // First entry found wins — with only one state to consider
+      // (Pending), there's no priority to resolve between DAGs.
+      if (!out.has(n.agent)) out.set(n.agent, e.kind);
+    }
+  }
+  return out;
+}
+
+// ─── transients ─────────────────────────────────────────────────────────────
 
 export function syncTransientsFromSnapshot(s) {
   transientsState.clear();
@@ -300,14 +344,9 @@ function buildContainerLi(c, node, opts) {
     askerCount, targetCount, agentQCount,
     url, containerBase, forgeBase, s,
   } = opts;
-  // A single `pending-running` class now covers the whole "has at
-  // least one badge" state — there's no more queued-but-not-running
-  // row to distinguish it from (see the pending-badge derivation
-  // comment in renderContainers), so the separate no-tint `pending`
-  // class from before that removal is gone rather than kept as a
-  // class that would now always co-occur with this one.
   const li = el('li', {
     class: 'container-row'
+      + (pending.length ? ' pending' : '')
       + (opRunning ? ' pending-running' : '')
       + (selected ? ' selected' : ''),
   });
@@ -583,15 +622,27 @@ export function renderContainers(s) {
     ));
   }
 
-  // No queue-summary banner for now — the previous version derived it
+  // Queue-summary banner: when the rebuild queue has active work,
-  // client-side from GET /api/jobq/graph, which mara flagged as
+  // show one compact at-a-glance line + a link to the full queue on the
-  // replacing one legacy DagView-shaped hack with another rather than
+  // C0R3 page. Replaces the old per-transient spinner list — the actual
-  // landing the real fix ("dont replace one legacy thing with another
+  // running step is already visible per-agent on each card (transient +
-  // ... split it into what can and cannot be done now"). Comes back once
+  // in-flight-queue badges), so the top of the tab only needs the summary.
-  // the dedicated rollup endpoint exists, reading that directly. Until
+  const activeQueue = rebuildQueueState.filter(
-  // then the queue's actual state is still visible per-agent on each
+    (e) => e.state === 'Pending' || e.state === 'Running',
-  // card via the transient pills; only the hive-wide at-a-glance summary
+  );
-  // is missing.
+  if (activeQueue.length) {
+    const running = activeQueue.filter((e) => e.state === 'Running').length;
+    const queued = activeQueue.length - running;
+    const parts = [];
+    if (running) parts.push(`${running} running`);
+    if (queued) parts.push(`${queued} queued`);
+    root.append(el('div', { class: 'queue-summary' },
+      el('span', { class: 'glyph spinner' }, '◐'), ' ',
+      el('strong', {}, 'build queue'), ' — ',
+      parts.join(' · '), ' ',
+      el('a', { class: 'queue-summary-link', href: '/builds.html' }, 'view queue →'),
+    ));
+  }
 
   if (!containers.length && !transientsState.size) {
     root.append(el('p', { class: 'empty' }, 'no managed containers'));
@@ -621,6 +672,10 @@ export function renderContainers(s) {
   const forgeBase = (s && s.forge_public_url) || null;
   const ul = existingUl ?? el('ul', { class: 'containers' });
   const tree = buildAgentTree(containers);
+  // Queued (not-yet-running) ops per agent name — see
+  // docs/web-ui.md::Container row for why this only covers the
+  // Pending case now (Running is fully covered by transientsState).
+  const queuedOps = queuedOpsByAgent();
 
   // Build the ordered list of <li> elements, reusing cached rows
   // whose displayed state hasn't changed.
@@ -634,17 +689,20 @@ export function renderContainers(s) {
     const containerBase = gatewayLinks
       ? `/agent/${encodeURIComponent(c.name)}`
       : `http://${hostname}:${c.port}`;
-    // Pending-badge derivation: transients only — "what is running,"
+    // Pending-badge derivation: an agent's transients win outright when
-    // full stop (mara, on review: "remove the per agent pending stuff -
+    // any exist (rendered one badge per pill — mara: "show all running
-    // only show what is running"; rendered one badge per pill — mara,
+    // nodes that name the agent"), the rebuild-queue's queued-only
-    // earlier: "show all running nodes that name the agent"). See
+    // fallback otherwise. See docs/web-ui.md::Container row.
-    // docs/web-ui.md::Container row.
     const transientKindsMap = transientsState.get(c.name);
     // Sorted for stable badge order across renders — Map iteration
     // order is insertion order, which shifts as pills clear/re-add.
-    const pending = transientKindsMap
+    const transientKinds = transientKindsMap
       ? Array.from(transientKindsMap.keys()).sort() : [];
-    const opRunning = pending.length > 0;
+    const queuedKind = transientKinds.length === 0 ? queuedOps.get(c.name) : null;
+    const pending = transientKinds.length > 0
+      ? transientKinds
+      : (queuedKind ? [queuedKind] : []);
+    const opRunning = transientKinds.length > 0;
     const selected = selectionState.has(c.name);
     // Pending questions where this agent is the asker (awaiting an
     // answer) or the target (owes a reply). Derived live from

frontend/packages/dashboard/src/tabs.js

@@ -42,8 +42,8 @@ import {
   renderQuestions, activeQuestionCount,
 } from './call.js';
 import {
-  syncTransientsFromSnapshot,
+  syncRebuildQueueFromSnapshot, syncTransientsFromSnapshot,
-  applyContainerStateChanged, applyContainerRemoved,
+  applyRebuildQueueChanged, applyContainerStateChanged, applyContainerRemoved,
   applyTransientSet, applyTransientCleared,
   renderContainers, renderContainersFromState,
   renderSelectionBar, renderPeerHives,
@@ -234,6 +234,9 @@ window.marked = marked;
       // `transientsState` + `containersState`, not from `s.*`).
       syncTransientsFromSnapshot(s);
       syncContainersFromSnapshot(s);
+      // Rebuild-queue state feeds the SW4RM agent-card badges
+      // (inFlightOpsByAgent); its own panel now lives on /core.html.
+      syncRebuildQueueFromSnapshot(s);
       renderContainers(s);
       // Sync the derived approvals + questions stores from the
       // snapshot, then render. Live `*_added` / `*_resolved` events
@@ -304,10 +307,8 @@ window.marked = marked;
     container_removed:       applyContainerRemoved,
     // tombstones_changed / meta_inputs_changed / meta_update_running are
     // handled on /core.html now (the SYST3M panels moved there).
-    // rebuild_queue_changed: this page no longer subscribes to it at
+    // rebuild_queue_changed stays: it refreshes the SW4RM badges.
-    // all (see swarm.js's transients-section comment) — the SW4RM tab
+    rebuild_queue_changed:   applyRebuildQueueChanged,
-    // has nothing left that reads it, unlike /builds.html which still
-    // does (its own separate subscription).
     schedules_changed:       applySchedulesChanged,
     capabilities_changed:    applyCapabilitiesChanged,
     tool_groups_changed:     applyToolGroupsChanged,
@@ -329,7 +330,7 @@ window.marked = marked;
     const es = openStream(
       '/api/dashboard/stream?kinds=sent,approval_added,approval_resolved,' +
         'question_added,question_resolved,transient_set,transient_cleared,' +
-        'container_state_changed,container_removed,' +
+        'container_state_changed,container_removed,rebuild_queue_changed,' +
         'schedules_changed,capabilities_changed,tool_groups_changed',
     );
     es.onmessage = (e) => {