fix(desktop): make Agents indicator match the Spawn-tree panel

The status-bar "Agents" item conflated three unrelated signals — running
subagents (aggregated across all sessions), in-flight session turns, and
failed background *system* actions (gateway restarts, toolset installs,
computer-use grants via $desktopActionTasks/preview restart) — yet
clicking it opens AgentsView, which renders only subagents. A failed
gateway restart therefore showed "Agents (1 Failed)" over an empty
"No live subagents" tree. AgentsView also filtered to the active session,
so a subagent running in a background session showed "Agents N running"
with nothing in the tree (the desync reported in #49808).

Unify the scope both surfaces speak:
- AgentsView aggregates subagents across every session (salvages #49819).
- The indicator's running/failed counts come from subagents only
  (aggregated), never background system actions — those keep their own
  surfaces in settings / command center.

So "Agents (N …)" now always points at a populated Spawn tree.

Supersedes #49819. Fixes #49808.

apps/desktop/src/store/subagents.test.ts

@@ -3,8 +3,10 @@ import { beforeEach, describe, expect, it, vi } from 'vitest'
 import {
   $subagentsBySession,
   activeSubagentCount,
+  allSubagents,
   buildSubagentTree,
   clearSessionSubagents,
+  failedSubagentCount,
   pruneDelegateFallbackSubagents,
   upsertSubagent
 } from './subagents'
@@ -99,6 +101,27 @@ describe('subagent store', () => {
     expect(listFor('s1').map(item => item.id)).toEqual(['sa-0-xyz'])
   })
 
+  // Contract: the status-bar "Agents" indicator and the Spawn-tree panel read
+  // the same scope — every session's subagents — so a count can never point at
+  // an empty tree (the desync behind "Agents (N)" vs "No live subagents").
+  it('counts running/failed across every session, matching the aggregated tree', () => {
+    upsertSubagent('s1', { goal: 'a', status: 'running', subagent_id: 'a', task_index: 0 })
+    upsertSubagent('s1', { goal: 'b', status: 'failed', subagent_id: 'b', task_index: 1 })
+    upsertSubagent('s2', { goal: 'c', status: 'running', subagent_id: 'c', task_index: 0 })
+    upsertSubagent('s2', { goal: 'd', status: 'failed', subagent_id: 'd', task_index: 1 })
+
+    const flat = allSubagents($subagentsBySession.get())
+    const indicatorRunning = Object.values($subagentsBySession.get()).reduce((n, l) => n + activeSubagentCount(l), 0)
+    const indicatorFailed = Object.values($subagentsBySession.get()).reduce((n, l) => n + failedSubagentCount(l), 0)
+    const tree = buildSubagentTree(flat)
+
+    // The active-session-only filter would have reported 1/1 here, not 2/2.
+    expect(indicatorRunning).toBe(2)
+    expect(indicatorFailed).toBe(2)
+    expect(tree).toHaveLength(4)
+    expect(indicatorRunning + indicatorFailed).toBe(tree.length)
+  })
+
   it('clears one session without touching another', () => {
     upsertSubagent('s1', { goal: 'one', status: 'running', subagent_id: 'a1', task_index: 0 })
     upsertSubagent('s2', { goal: 'two', status: 'running', subagent_id: 'a2', task_index: 0 })

apps/desktop/src/store/subagents.ts

@@ -261,3 +261,10 @@ export function buildSubagentTree(items: readonly SubagentProgress[]): SubagentN
 
 export const activeSubagentCount = (items: readonly SubagentProgress[]) =>
   items.filter(item => item.status === 'queued' || item.status === 'running').length
+
+export const failedSubagentCount = (items: readonly SubagentProgress[]) =>
+  items.filter(item => item.status === 'failed' || item.status === 'interrupted').length
+
+/** Flatten every session's subagents — the scope the Spawn-tree panel and the
+ *  status-bar indicator must agree on. */
+export const allSubagents = (bySession: Record<string, SubagentProgress[]>) => Object.values(bySession).flat()