diff --git a/frontend/web/src/components/intelligent_eval/ConfigSnapshots.test.tsx b/frontend/web/src/components/intelligent_eval/ConfigSnapshots.test.tsx
new file mode 100644
index 0000000..5597dfa
--- /dev/null
+++ b/frontend/web/src/components/intelligent_eval/ConfigSnapshots.test.tsx
@@ -0,0 +1,116 @@
+import { cleanup, fireEvent, render, screen, waitFor } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import type { ConfigSnapshot } from '../../api'
+import ConfigSnapshots from './ConfigSnapshots'
+
+const listConfigSnapshots = vi.fn()
+const getConfigSnapshot = vi.fn()
+const compareConfigSnapshots = vi.fn()
+const downloadJson = vi.fn()
+
+vi.mock('../../api', () => ({
+  intelligentEvalsApi: {
+    listConfigSnapshots: (...args: unknown[]) => listConfigSnapshots(...args),
+    getConfigSnapshot: (...args: unknown[]) => getConfigSnapshot(...args),
+    compareConfigSnapshots: (...args: unknown[]) => compareConfigSnapshots(...args),
+  },
+}))
+
+vi.mock('../../utils/download', () => ({
+  downloadJson: (...args: unknown[]) => downloadJson(...args),
+}))
+
+function makeSnapshot(id: string, type: ConfigSnapshot['snapshot_type'] = 'created'): ConfigSnapshot {
+  return {
+    id,
+    eval_id: 'eval-1',
+    snapshot_type: type,
+    goal: `评估目标 ${id}`,
+    seeds: {},
+    intent: '',
+    role_description: '',
+    time_window_hours: 24,
+    plan: null,
+    created_at: '2026-01-01T08:00:00Z',
+    created_by: 'openclaw',
+  }
+}
+
+beforeEach(() => {
+  vi.clearAllMocks()
+})
+
+afterEach(() => {
+  cleanup()
+})
+
+describe('ConfigSnapshots', () => {
+  it('lists snapshots with type labels and shows empty state when none', async () => {
+    listConfigSnapshots.mockResolvedValue({ data: { snapshots: [makeSnapshot('snap-1'), makeSnapshot('snap-2', 'plan_submitted')] } })
+    render(<ConfigSnapshots evalId="eval-1" />)
+
+    await waitFor(() => expect(listConfigSnapshots).toHaveBeenCalledWith('eval-1'))
+    expect(await screen.findByText('创建')).toBeInTheDocument()
+    expect(screen.getByText('计划提交')).toBeInTheDocument()
+    expect(screen.getAllByText('openclaw').length).toBe(2)
+
+    cleanup()
+    listConfigSnapshots.mockResolvedValue({ data: { snapshots: [] } })
+    render(<ConfigSnapshots evalId="eval-1" />)
+    expect(await screen.findByText('暂无配置快照')).toBeInTheDocument()
+  })
+
+  it('opens snapshot detail and returns to the list', async () => {
+    const snapshot = makeSnapshot('snap-1')
+    listConfigSnapshots.mockResolvedValue({ data: { snapshots: [snapshot] } })
+    getConfigSnapshot.mockResolvedValue({ data: snapshot })
+    render(<ConfigSnapshots evalId="eval-1" />)
+
+    fireEvent.click(await screen.findByText(/查\s*看/))
+    await waitFor(() => expect(getConfigSnapshot).toHaveBeenCalledWith('eval-1', 'snap-1'))
+    expect(await screen.findByText('快照详情')).toBeInTheDocument()
+    expect(screen.getByText('评估目标 snap-1')).toBeInTheDocument()
+    expect(screen.getByText('24 小时')).toBeInTheDocument()
+
+    fireEvent.click(screen.getByText('返回'))
+    expect(await screen.findByText('配置历史')).toBeInTheDocument()
+  })
+
+  it('exports a snapshot as JSON', async () => {
+    listConfigSnapshots.mockResolvedValue({ data: { snapshots: [makeSnapshot('snap-12345678')] } })
+    render(<ConfigSnapshots evalId="eval-1" />)
+
+    fireEvent.click(await screen.findByText(/导\s*出/))
+    expect(downloadJson).toHaveBeenCalledWith(
+      expect.objectContaining({ id: 'snap-12345678' }),
+      'config-snapshot-snap-123.json',
+    )
+  })
+
+  it('compares two selected snapshots and renders differences', async () => {
+    listConfigSnapshots.mockResolvedValue({
+      data: { snapshots: [makeSnapshot('snap-1'), makeSnapshot('snap-2', 'config_updated')] },
+    })
+    compareConfigSnapshots.mockResolvedValue({
+      data: {
+        snapshot_1: { id: 'snap-1', snapshot_type: 'created', created_at: '2026-01-01T08:00:00Z' },
+        snapshot_2: { id: 'snap-2', snapshot_type: 'config_updated', created_at: '2026-01-02T08:00:00Z' },
+        differences: { goal: { old: '旧目标', new: '新目标' } },
+      },
+    })
+    render(<ConfigSnapshots evalId="eval-1" />)
+
+    await screen.findByText('创建')
+    // 第 0 个是表头全选框，行内勾选从索引 1 开始
+    const checkboxes = screen.getAllByRole('checkbox')
+    fireEvent.click(checkboxes[1])
+    fireEvent.click(checkboxes[2])
+    fireEvent.click(screen.getByText(/对比选中/))
+
+    await waitFor(() => expect(compareConfigSnapshots).toHaveBeenCalledWith('eval-1', 'snap-1', 'snap-2'))
+    expect(await screen.findByText('快照对比')).toBeInTheDocument()
+    expect(screen.getByText('goal')).toBeInTheDocument()
+    expect(screen.getByText('旧目标')).toBeInTheDocument()
+    expect(screen.getByText('新目标')).toBeInTheDocument()
+  })
+})
diff --git a/frontend/web/src/components/intelligent_eval/DecisionProcess.test.tsx b/frontend/web/src/components/intelligent_eval/DecisionProcess.test.tsx
new file mode 100644
index 0000000..50adc37
--- /dev/null
+++ b/frontend/web/src/components/intelligent_eval/DecisionProcess.test.tsx
@@ -0,0 +1,90 @@
+import { cleanup, fireEvent, render, screen } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import type { DecisionLog } from '../../api'
+import type { ReadSlot } from '../../read/readResource'
+import DecisionProcess from './DecisionProcess'
+
+const downloadJson = vi.fn()
+
+vi.mock('../../utils/download', () => ({
+  downloadJson: (...args: unknown[]) => downloadJson(...args),
+}))
+
+function makeLog(i: number, type: DecisionLog['decision_type'] = 'execute_session'): DecisionLog {
+  return {
+    id: `log-${i}`,
+    eval_id: 'eval-1',
+    decision_type: type,
+    reason: `时段欠账 ${i}`,
+    context: { debt: i },
+    cron_id: `cron-${i}-abcdef`,
+    created_at: `2026-01-01T08:${String(i % 60).padStart(2, '0')}:00Z`,
+  }
+}
+
+function readySlot(logs: DecisionLog[]): ReadSlot<DecisionLog[] | null> {
+  return { phase: 'ready', value: logs, error: null }
+}
+
+beforeEach(() => {
+  vi.clearAllMocks()
+})
+
+afterEach(() => {
+  cleanup()
+})
+
+describe('DecisionProcess', () => {
+  it('shows loading placeholder while the slot is loading', () => {
+    const slot: ReadSlot<DecisionLog[] | null> = { phase: 'loading', value: null, error: null }
+    render(<DecisionProcess evalId="eval-1" logs={slot} onReload={() => {}} />)
+
+    expect(screen.getByText('加载中...')).toBeInTheDocument()
+  })
+
+  it('shows empty state when there are no logs', () => {
+    render(<DecisionProcess evalId="eval-1" logs={readySlot([])} onReload={() => {}} />)
+
+    expect(screen.getAllByText('暂无决策日志').length).toBeGreaterThanOrEqual(1)
+  })
+
+  it('renders log reasons in the timeline and the table', () => {
+    render(<DecisionProcess evalId="eval-1" logs={readySlot([makeLog(1), makeLog(2, 'wait')])} onReload={() => {}} />)
+
+    expect(screen.getAllByText('时段欠账 1').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getAllByText('时段欠账 2').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getByText(/共 2 条/)).toBeInTheDocument()
+  })
+
+  it('expands a row to show decision context', async () => {
+    render(<DecisionProcess evalId="eval-1" logs={readySlot([makeLog(3)])} onReload={() => {}} />)
+
+    expect(screen.getAllByText('时段欠账 3').length).toBe(2)
+    expect(screen.queryByText('决策上下文')).not.toBeInTheDocument()
+    // antd 会在两个汉字的按钮文案中间插空格（详 情）
+    fireEvent.click(screen.getByText(/详\s*情/))
+    expect(await screen.findByText('决策上下文')).toBeInTheDocument()
+    expect(screen.getByText(/"debt": 3/)).toBeInTheDocument()
+  })
+
+  it('triggers onReload from the refresh button', () => {
+    const onReload = vi.fn()
+    render(<DecisionProcess evalId="eval-1" logs={readySlot([])} onReload={onReload} />)
+
+    fireEvent.click(screen.getByText('刷新'))
+    expect(onReload).toHaveBeenCalledTimes(1)
+  })
+
+  it('exports logs as JSON and skips export when empty', () => {
+    const { unmount } = render(
+      <DecisionProcess evalId="eval-12345678" logs={readySlot([makeLog(1)])} onReload={() => {}} />,
+    )
+    fireEvent.click(screen.getByText(/导\s*出/))
+    expect(downloadJson).toHaveBeenCalledWith(expect.any(Array), 'decision-logs-eval-123.json')
+    unmount()
+
+    render(<DecisionProcess evalId="eval-1" logs={readySlot([])} onReload={() => {}} />)
+    fireEvent.click(screen.getByText(/导\s*出/))
+    expect(downloadJson).toHaveBeenCalledTimes(1)
+  })
+})
diff --git a/frontend/web/src/components/intelligent_eval/EvalOverview.test.tsx b/frontend/web/src/components/intelligent_eval/EvalOverview.test.tsx
new file mode 100644
index 0000000..1b9a603
--- /dev/null
+++ b/frontend/web/src/components/intelligent_eval/EvalOverview.test.tsx
@@ -0,0 +1,129 @@
+import { cleanup, render, screen } from '@testing-library/react'
+import { afterEach, describe, expect, it } from 'vitest'
+import type { IntelligentEval, IntelligentEvalSession } from '../../api'
+import EvalOverview from './EvalOverview'
+
+function makeEval(overrides: Partial<IntelligentEval> = {}): IntelligentEval {
+  return {
+    id: 'eval-1',
+    name: '24h 服务质量评估',
+    target_id: 't-1',
+    status: 'pending_approval',
+    goal: '评估服务态度与专业度',
+    seeds: {},
+    intent: '重点关注价保场景',
+    role_description: '客服数字员工',
+    plan: null,
+    plan_feedback: null,
+    time_window_hours: 24,
+    report: null,
+    created_at: '2026-01-01T00:00:00Z',
+    updated_at: null,
+    started_at: null,
+    completed_at: null,
+    session_count: 4,
+    completed_sessions: 0,
+    sessions: [],
+    ...overrides,
+  }
+}
+
+function makeSession(overrides: Partial<IntelligentEvalSession> = {}): IntelligentEvalSession {
+  return {
+    id: 's-1',
+    eval_id: 'eval-1',
+    target_id: 't-1',
+    persona: { name: '老客户' },
+    goal: '完成退货',
+    dimension: '售后',
+    status: 'completed',
+    verdict: null,
+    turn_count: 5,
+    created_at: '2026-01-01T08:10:00Z',
+    closed_at: '2026-01-01T08:25:00Z',
+    ...overrides,
+  }
+}
+
+afterEach(() => {
+  cleanup()
+})
+
+describe('EvalOverview', () => {
+  it('renders basic info: target, status tag, time window and session progress', () => {
+    render(<EvalOverview ev={makeEval()} targetName="tutu 客服通道" />)
+
+    expect(screen.getByText('tutu 客服通道')).toBeInTheDocument()
+    expect(screen.getByText('待审批')).toBeInTheDocument()
+    expect(screen.getByText('24 小时')).toBeInTheDocument()
+    expect(screen.getByText('0/4')).toBeInTheDocument()
+    expect(screen.getByText('评估服务态度与专业度')).toBeInTheDocument()
+    expect(screen.getByText('重点关注价保场景')).toBeInTheDocument()
+  })
+
+  it('shows planning hint while planning without feedback', () => {
+    render(<EvalOverview ev={makeEval({ status: 'planning' })} targetName="t" />)
+
+    expect(screen.getByText(/OpenClaw 正在规划/)).toBeInTheDocument()
+    expect(screen.queryByText('粗计划')).not.toBeInTheDocument()
+  })
+
+  it('shows rejection alert when plan was sent back', () => {
+    render(<EvalOverview ev={makeEval({ status: 'planning', plan_feedback: '维度太粗，请细化' })} targetName="t" />)
+
+    expect(screen.getByText(/计划已打回/)).toBeInTheDocument()
+    expect(screen.getByText('维度太粗，请细化')).toBeInTheDocument()
+  })
+
+  it('renders the plan summary for pending_approval', () => {
+    const ev = makeEval({
+      plan: {
+        dimensions: ['服务态度', '专业度'],
+        virtual_users: [{ persona: { name: '新客户' }, goal: '咨询价保' }],
+        time_distribution: [{ time_slot: '0-1h', sessions: 2, scenario: '早高峰咨询' }],
+        estimated_sessions: 4,
+        budget: { max_turns_per_session: 12, total_max_turns: 48 },
+        completion_criteria: '全部时段完成',
+      },
+    })
+    render(<EvalOverview ev={ev} targetName="t" />)
+
+    expect(screen.getByText('粗计划')).toBeInTheDocument()
+    expect(screen.getByText('服务态度')).toBeInTheDocument()
+    expect(screen.getByText('专业度')).toBeInTheDocument()
+    expect(screen.getByText('新客户')).toBeInTheDocument()
+    // 目标与 persona 在同一个 div 里（<b>新客户</b> · 咨询价保）
+    expect(screen.getByText((content) => content.includes('咨询价保'))).toBeInTheDocument()
+    expect(screen.getByText('0-1h')).toBeInTheDocument()
+    expect(screen.getByText('全部时段完成')).toBeInTheDocument()
+  })
+
+  it('renders session progress rows only while executing or completed', () => {
+    const sessions = [makeSession()]
+    const { unmount } = render(
+      <EvalOverview
+        ev={makeEval({ status: 'executing', sessions, completed_sessions: 1, session_count: 1 })}
+        targetName="t"
+      />,
+    )
+    expect(screen.getByText(/会话进度（1\/1）/)).toBeInTheDocument()
+    expect(screen.getByText('老客户')).toBeInTheDocument()
+    expect(screen.getByText('完成退货')).toBeInTheDocument()
+    expect(screen.getByText('售后')).toBeInTheDocument()
+    unmount()
+
+    render(<EvalOverview ev={makeEval({ status: 'pending_approval', sessions })} targetName="t" />)
+    expect(screen.queryByText(/会话进度（/)).not.toBeInTheDocument()
+  })
+
+  it('shows seeds JSON when provided and a dash when empty', () => {
+    const { unmount } = render(
+      <EvalOverview ev={makeEval({ seeds: { questions: ['怎么退货'] } })} targetName="t" />,
+    )
+    expect(screen.getByText(/怎么退货/)).toBeInTheDocument()
+    unmount()
+
+    render(<EvalOverview ev={makeEval()} targetName="t" />)
+    expect(screen.getByText('已显示全部内容')).toBeInTheDocument()
+  })
+})
diff --git a/frontend/web/src/components/intelligent_eval/EvalReport.test.tsx b/frontend/web/src/components/intelligent_eval/EvalReport.test.tsx
new file mode 100644
index 0000000..15da14a
--- /dev/null
+++ b/frontend/web/src/components/intelligent_eval/EvalReport.test.tsx
@@ -0,0 +1,139 @@
+import { cleanup, fireEvent, render, screen, waitFor } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import type { IntelligentEval, IntelligentEvalReport } from '../../api'
+import EvalReport from './EvalReport'
+
+const downloadReportMarkdown = vi.fn()
+const listMessages = vi.fn()
+
+vi.mock('../../api', () => ({
+  intelligentEvalsApi: {
+    downloadReportMarkdown: (...args: unknown[]) => downloadReportMarkdown(...args),
+    listMessages: (...args: unknown[]) => listMessages(...args),
+  },
+}))
+
+function makeReport(overrides: Partial<IntelligentEvalReport> = {}): IntelligentEvalReport {
+  return {
+    summary: '整体服务态度良好，价保场景存在流程遗漏',
+    scores: { overall: 4.2, dimensions: { 服务态度: 4.5, 专业度: 3.2 } },
+    findings: [
+      {
+        issue: '价保流程未主动告知凭证',
+        severity: 'high',
+        dimension: '专业度',
+        evidence: [{ session_id: 'sess-aaaaaaaa', turn_index: 2, user_said: '需要什么凭证', assistant_replied: '您好' }],
+        suggestion: '补充 SOP 提示',
+        related_sop: 'SOP-12',
+      },
+      {
+        issue: '回复偏慢',
+        severity: 'medium',
+        dimension: '响应时效',
+        evidence: [],
+      },
+    ],
+    highlights: [{ description: '语气友好', dimension: '服务态度' }],
+    priority_recommendations: ['优先整改价保流程'],
+    ...overrides,
+  }
+}
+
+function makeEval(overrides: Partial<IntelligentEval> = {}): IntelligentEval {
+  return {
+    id: 'eval-1',
+    name: '24h 评估',
+    target_id: 't-1',
+    status: 'completed',
+    goal: 'goal',
+    seeds: {},
+    intent: '',
+    role_description: '',
+    plan: null,
+    plan_feedback: null,
+    time_window_hours: 24,
+    report: makeReport(),
+    created_at: null,
+    updated_at: null,
+    started_at: null,
+    completed_at: '2026-01-02T00:00:00Z',
+    session_count: 1,
+    completed_sessions: 1,
+    sessions: [
+      {
+        id: 's-1',
+        eval_id: 'eval-1',
+        target_id: 't-1',
+        persona: { name: '老客户' },
+        goal: '完成退货',
+        dimension: '售后',
+        status: 'completed',
+        verdict: null,
+        turn_count: 5,
+        created_at: '2026-01-01T08:10:00Z',
+        closed_at: '2026-01-01T08:25:00Z',
+      },
+    ],
+    ...overrides,
+  }
+}
+
+beforeEach(() => {
+  vi.clearAllMocks()
+})
+
+afterEach(() => {
+  cleanup()
+})
+
+describe('EvalReport', () => {
+  it('shows empty states when there is no report and no sessions', () => {
+    render(<EvalReport ev={makeEval({ report: null, sessions: [] })} />)
+
+    expect(screen.getByText('暂无报告')).toBeInTheDocument()
+    expect(screen.getByText('暂无会话')).toBeInTheDocument()
+  })
+
+  it('renders summary cards, dimension scores, findings, highlights and recommendations', () => {
+    render(<EvalReport ev={makeEval()} />)
+
+    expect(screen.getByText('4.2')).toBeInTheDocument()
+    expect(screen.getByText('问题发现（2）')).toBeInTheDocument()
+    expect(screen.getAllByText('服务态度').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getByText('4.5')).toBeInTheDocument()
+    expect(screen.getByText('3.2')).toBeInTheDocument()
+    expect(screen.getByText('整体服务态度良好，价保场景存在流程遗漏')).toBeInTheDocument()
+    expect(screen.getByText('价保流程未主动告知凭证')).toBeInTheDocument()
+    expect(screen.getByText('回复偏慢')).toBeInTheDocument()
+    expect(screen.getByText('语气友好')).toBeInTheDocument()
+    expect(screen.getByText('优先整改价保流程')).toBeInTheDocument()
+    expect(document.body.textContent).toContain('高 1 · 中 1 · 低 0')
+  })
+
+  it('expanding a session panel loads its messages', async () => {
+    listMessages.mockResolvedValue({
+      data: {
+        messages: [
+          { id: 'm-1', session_id: 's-1', role: 'user', content: '我想退货', latency_ms: null, created_at: null },
+          { id: 'm-2', session_id: 's-1', role: 'assistant', content: '好的，请提供订单号', latency_ms: 320, created_at: null },
+        ],
+      },
+    })
+    render(<EvalReport ev={makeEval()} />)
+
+    expect(listMessages).not.toHaveBeenCalled()
+    fireEvent.click(screen.getByText('老客户'))
+    await waitFor(() => expect(listMessages).toHaveBeenCalledWith('eval-1', 's-1'))
+    expect(await screen.findByText('我想退货')).toBeInTheDocument()
+    expect(screen.getByText('好的，请提供订单号')).toBeInTheDocument()
+    expect(screen.getByText('320ms')).toBeInTheDocument()
+  })
+
+  it('exports the markdown report', async () => {
+    downloadReportMarkdown.mockResolvedValue(undefined)
+    render(<EvalReport ev={makeEval()} />)
+
+    fireEvent.click(screen.getByText(/导出 Markdown/))
+    await waitFor(() => expect(downloadReportMarkdown).toHaveBeenCalledWith('eval-1'))
+  })
+})
diff --git a/frontend/web/src/components/intelligent_eval/TaskQueueMonitor.test.tsx b/frontend/web/src/components/intelligent_eval/TaskQueueMonitor.test.tsx
new file mode 100644
index 0000000..3bc7735
--- /dev/null
+++ b/frontend/web/src/components/intelligent_eval/TaskQueueMonitor.test.tsx
@@ -0,0 +1,126 @@
+import { cleanup, fireEvent, render, screen, waitFor } from '@testing-library/react'
+import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'
+import type { TaskQueueItem, TaskQueueList } from '../../api'
+import TaskQueueMonitor from './TaskQueueMonitor'
+
+const listTasks = vi.fn()
+
+vi.mock('../../api', () => ({
+  intelligentEvalsApi: {
+    listTasks: (...args: unknown[]) => listTasks(...args),
+  },
+}))
+
+function makeTask(id: string, status: TaskQueueItem['status'], overrides: Partial<TaskQueueItem> = {}): TaskQueueItem {
+  return {
+    id,
+    eval_id: `eval-${id}`,
+    eval_name: `评估 ${id}`,
+    eval_status: 'executing',
+    status,
+    priority: 1,
+    reason: `时段欠账 ${id}`,
+    assigned_cron_id: null,
+    assigned_at: null,
+    completed_at: status === 'completed' ? '2026-01-01T09:00:00Z' : null,
+    created_at: '2026-01-01T08:00:00Z',
+    updated_at: null,
+    error: null,
+    ...overrides,
+  }
+}
+
+function makeList(tasks: TaskQueueItem[], stats: TaskQueueList['stats']): TaskQueueList {
+  return { tasks, stats }
+}
+
+beforeEach(() => {
+  vi.clearAllMocks()
+})
+
+afterEach(() => {
+  cleanup()
+  vi.useRealTimers()
+})
+
+describe('TaskQueueMonitor', () => {
+  it('renders stat chips from stats without the derived unresolved chip', async () => {
+    listTasks.mockResolvedValue({
+      data: makeList([makeTask('1', 'pending')], { pending: 1, assigned: 2, completed: 3, failed: 4, unresolved: 3 }),
+    })
+    render(<TaskQueueMonitor />)
+
+    await waitFor(() => expect(listTasks).toHaveBeenCalledTimes(1))
+    expect(await screen.findByText('全部任务')).toBeInTheDocument()
+    // 状态文案同时出现在 chip 与表格 Tag 中
+    expect(screen.getAllByText('待认领').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getAllByText('执行中').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getAllByText('已完成').length).toBeGreaterThanOrEqual(1)
+    expect(screen.getAllByText('失败').length).toBeGreaterThanOrEqual(1)
+    // 全部 = 四态之和，不含派生的 unresolved
+    expect(screen.getByText('10')).toBeInTheDocument()
+    expect(screen.queryByText('unresolved')).not.toBeInTheDocument()
+  })
+
+  it('renders task rows with eval name and queue status', async () => {
+    listTasks.mockResolvedValue({
+      data: makeList(
+        [
+          makeTask('1', 'completed'),
+          makeTask('2', 'failed', { error: 'worker 超时' }),
+        ],
+        { pending: 0, assigned: 0, completed: 1, failed: 1, unresolved: 0 },
+      ),
+    })
+    render(<TaskQueueMonitor />)
+
+    expect(await screen.findByText('评估 1')).toBeInTheDocument()
+    expect(screen.getByText('评估 2')).toBeInTheDocument()
+    expect(screen.getByText('时段欠账 1')).toBeInTheDocument()
+    // 失败任务的完成时间列显示 失败（悬浮提示里是错误详情）
+    expect(screen.getAllByText('失败').length).toBeGreaterThanOrEqual(1)
+  })
+
+  it('filters rows by the clicked stat chip', async () => {
+    listTasks.mockResolvedValue({
+      data: makeList(
+        [makeTask('1', 'pending'), makeTask('2', 'completed')],
+        { pending: 1, assigned: 0, completed: 1, failed: 0, unresolved: 1 },
+      ),
+    })
+    render(<TaskQueueMonitor />)
+
+    await screen.findByText('评估 1')
+    expect(screen.getByText('评估 2')).toBeInTheDocument()
+    // stat-bar 先于表格渲染，第一个 待认领 是筛选 chip
+    fireEvent.click(screen.getAllByText('待认领')[0])
+    expect(screen.getByText('评估 1')).toBeInTheDocument()
+    expect(screen.queryByText('评估 2')).not.toBeInTheDocument()
+  })
+
+  it('shows empty state when there are no tasks', async () => {
+    listTasks.mockResolvedValue({
+      data: makeList([], { pending: 0, assigned: 0, completed: 0, failed: 0, unresolved: 0 }),
+    })
+    render(<TaskQueueMonitor />)
+
+    expect(await screen.findByText('暂无任务')).toBeInTheDocument()
+  })
+
+  it('polls every 5 seconds and reloads on demand', async () => {
+    vi.useFakeTimers()
+    listTasks.mockResolvedValue({
+      data: makeList([], { pending: 0, assigned: 0, completed: 0, failed: 0, unresolved: 0 }),
+    })
+    render(<TaskQueueMonitor />)
+
+    await vi.advanceTimersByTimeAsync(0)
+    expect(listTasks).toHaveBeenCalledTimes(1)
+    await vi.advanceTimersByTimeAsync(5000)
+    expect(listTasks).toHaveBeenCalledTimes(2)
+    // 手动刷新按钮
+    fireEvent.click(screen.getByRole('button'))
+    await vi.advanceTimersByTimeAsync(0)
+    expect(listTasks).toHaveBeenCalledTimes(3)
+  })
+})