feat(intelligent-eval): paginate the eval list

GET /api/intelligent-evals 支持 page/page_size（默认不传仍返回全部，向后兼容）：
repository 加 count/list_page，lifecycle 加 list_evals_page，router 返回 total。
前端服务端分页：useIntelligentEvalRead 接 page/pageSize，list 存 total，
IntelligentEvals 表格 showSizeChanger + 页码切换重新加载；5s 轮询保持当前页。
测试：+3 后端分页 + hook 页码透传/总数断言，895 passed，vitest 19 passed

backend/agenteval/intelligent_eval/lifecycle.py

@@ -218,6 +218,12 @@ def list_evals(session: Session) -> list[IntelligentEval]:
     return IntelligentEvalRepository(session).list_all()
 
 
+def list_evals_page(session: Session, offset: int, limit: int) -> tuple[list[IntelligentEval], int]:
+    """Return one page of evals (newest first) plus the total count."""
+    repo = IntelligentEvalRepository(session)
+    return repo.list_page(offset, limit), repo.count()
+
+
 def _get_session_or_raise(repo: IntelligentEvalSessionRepository, session_id: str) -> IntelligentEvalSession:
     obj = repo.get(session_id)
     if obj is None:

backend/agenteval/intelligent_eval/repository.py

@@ -101,6 +101,20 @@ class IntelligentEvalRepository:
         statement = select(IntelligentEvalDB).order_by(IntelligentEvalDB.created_at.desc())
         return [self._from_db(r) for r in self.session.exec(statement).all()]
 
+    def list_page(self, offset: int, limit: int) -> list[IntelligentEval]:
+        """Return one page of evals (created_at desc) with the same ordering as ``list_all``."""
+        statement = (
+            select(IntelligentEvalDB)
+            .order_by(IntelligentEvalDB.created_at.desc())
+            .offset(offset)
+            .limit(limit)
+        )
+        return [self._from_db(r) for r in self.session.exec(statement).all()]
+
+    def count(self) -> int:
+        """Total number of evaluations (for pagination metadata)."""
+        return self.session.exec(select(func.count()).select_from(IntelligentEvalDB)).one()
+
     def get(self, eval_id: str) -> Optional[IntelligentEval]:
         db = self.session.get(IntelligentEvalDB, eval_id)
         return self._from_db(db) if db else None

backend/agenteval/web/routers/intelligent_evals.py

@@ -94,10 +94,29 @@ async def create_eval(request: CreateEvalRequest, session: Session = Depends(get
 
 
 @router.get("")
-async def list_evals(session: Session = Depends(get_db)) -> dict:
-    evals = lifecycle.list_evals(session)
+async def list_evals(
+    page: int | None = None,
+    page_size: int = 20,
+    session: Session = Depends(get_db),
+) -> dict:
+    """List intelligent evaluations, optionally paginated.
+
+    不传 ``page`` 时返回全部（向后兼容）；传 ``page``（从 1 起）时按
+    ``created_at`` 倒序分页，返回 ``total`` 供前端服务端分页。
+    """
     reader = IntelligentEvalReadModel(session)
+    if page is None:
+        evals = lifecycle.list_evals(session)
         return {"intelligent_evals": [item.model_dump(mode="json") for item in reader.list_items(evals)]}
+    page_size = max(1, min(page_size, 100))
+    offset = (max(1, page) - 1) * page_size
+    evals, total = lifecycle.list_evals_page(session, offset, page_size)
+    return {
+        "intelligent_evals": [item.model_dump(mode="json") for item in reader.list_items(evals)],
+        "total": total,
+        "page": page,
+        "page_size": page_size,
+    }
 
 
 @router.get("/tasks")

frontend/web/src/api.ts

@@ -785,7 +785,11 @@ export interface TaskQueueList {
 }
 
 export const intelligentEvalsApi = {
-  list: () => api.get<{ intelligent_evals: IntelligentEval[] }>('/intelligent-evals'),
+  list: (params?: { page?: number; page_size?: number }) =>
+    api.get<{ intelligent_evals: IntelligentEval[]; total?: number; page?: number; page_size?: number }>(
+      '/intelligent-evals',
+      { params },
+    ),
   listTasks: (params?: { status?: TaskQueueStatus; limit?: number }) =>
     api.get<TaskQueueList>('/intelligent-evals/tasks', { params }),
   get: (id: string) => api.get<IntelligentEval>(`/intelligent-evals/${id}`),

frontend/web/src/pages/IntelligentEvals.tsx

@@ -38,10 +38,12 @@ export default function IntelligentEvalsPage() {
   const [selectedId, setSelectedId] = useState<string | null>(null)
   const [createOpen, setCreateOpen] = useState(false)
   const [taskQueueOpen, setTaskQueueOpen] = useState(false)
+  const [page, setPage] = useState(1)
+  const [pageSize, setPageSize] = useState(20)
   const [submitting, setSubmitting] = useState(false)
   const [form] = Form.useForm<CreateFormValues>()
 
-  const { list, detail, reloadList, reloadDetail } = useIntelligentEvalRead(selectedId)
+  const { list, detail, reloadList, reloadDetail } = useIntelligentEvalRead(selectedId, undefined, page, pageSize)
   const evals = list.value
   const loading = list.phase === 'loading'
   const { data: targets } = useResource(
@@ -173,7 +175,18 @@ export default function IntelligentEvalsPage() {
           loading={loading}
           dataSource={evals ?? []}
           columns={columns}
-          pagination={(evals?.length ?? 0) > 20 ? { pageSize: 20, showTotal: (t) => `共 ${t} 个` } : false}
+          pagination={{
+            current: page,
+            pageSize,
+            total: list.total,
+            showSizeChanger: true,
+            pageSizeOptions: [10, 20, 50],
+            showTotal: (t) => `共 ${t} 个`,
+            onChange: (p, ps) => {
+              setPage(p)
+              setPageSize(ps)
+            },
+          }}
           locale={{ emptyText: <Empty description="还没有智能评估" /> }}
         />
       </div>

frontend/web/src/read/intelligentEval.test.ts

@@ -12,20 +12,25 @@ describe('intelligent evaluation read state', () => {
     const loading = intelligentEvalReadReducer(initialIntelligentEvalReadState, { type: 'list_requested' })
     expect(loading.list.phase).toBe('loading')
 
-    const ready = intelligentEvalReadReducer(loading, { type: 'list_succeeded', value: [evaluation] })
-    expect(ready.list).toEqual({ phase: 'ready', value: [evaluation], error: null })
+    const ready = intelligentEvalReadReducer(loading, {
+      type: 'list_succeeded',
+      value: [evaluation],
+      total: 42,
+    })
+    expect(ready.list).toEqual({ phase: 'ready', value: [evaluation], error: null, total: 42 })
   })
 
   it('keeps the last list snapshot during silent refresh and failure', () => {
     const ready: IntelligentEvalReadState = {
       ...initialIntelligentEvalReadState,
-      list: { phase: 'ready', value: [evaluation], error: null },
+      list: { phase: 'ready', value: [evaluation], error: null, total: 1 },
     }
     const refreshing = intelligentEvalReadReducer(ready, { type: 'list_requested', silent: true })
     expect(refreshing.list.phase).toBe('refreshing')
+    expect(refreshing.list.total).toBe(1)
 
     const afterFailure = intelligentEvalReadReducer(refreshing, { type: 'list_failed', error: '网络错误' })
-    expect(afterFailure.list).toEqual({ phase: 'ready', value: [evaluation], error: null })
+    expect(afterFailure.list).toEqual({ phase: 'ready', value: [evaluation], error: null, total: 1 })
   })
 
   it('surfaces an initial detail failure without an existing snapshot', () => {

frontend/web/src/read/intelligentEval.ts

@@ -9,7 +9,7 @@ export interface ReadSlot<T> {
 }
 
 export interface IntelligentEvalReadState {
-  list: ReadSlot<IntelligentEval[]>
+  list: ReadSlot<IntelligentEval[]> & { total: number }
   detail: ReadSlot<IntelligentEval | null> & {
     selectedId: string | null
     requestId: number
@@ -18,7 +18,7 @@ export interface IntelligentEvalReadState {
 
 export type IntelligentEvalReadAction =
   | { type: 'list_requested'; silent?: boolean }
-  | { type: 'list_succeeded'; value: IntelligentEval[] }
+  | { type: 'list_succeeded'; value: IntelligentEval[]; total: number }
   | { type: 'list_failed'; error: string }
   | { type: 'detail_cleared'; requestId: number }
   | { type: 'detail_requested'; id: string; requestId: number; silent?: boolean }
@@ -26,21 +26,28 @@ export type IntelligentEvalReadAction =
   | { type: 'detail_failed'; id: string; requestId: number; error: string }
 
 export interface IntelligentEvalReadAdapter {
-  list: () => Promise<IntelligentEval[]>
+  list: (page?: number, pageSize?: number) => Promise<{ items: IntelligentEval[]; total: number }>
   get: (id: string) => Promise<IntelligentEval>
 }
 
 export const intelligentEvalReadAdapter: IntelligentEvalReadAdapter = {
-  list: () => intelligentEvalsApi.list().then((response) => response.data.intelligent_evals),
+  list: (page, pageSize) =>
+    intelligentEvalsApi
+      .list(page != null ? { page, page_size: pageSize ?? 20 } : undefined)
+      .then((response) => ({
+        items: response.data.intelligent_evals,
+        // 兼容不分页响应：无 total 时以本页长度为总数
+        total: response.data.total ?? response.data.intelligent_evals.length,
+      })),
   get: (id) => intelligentEvalsApi.get(id).then((response) => response.data),
 }
 
 export const initialIntelligentEvalReadState: IntelligentEvalReadState = {
-  list: { phase: 'idle', value: [], error: null },
+  list: { phase: 'idle', value: [], error: null, total: 0 },
   detail: { phase: 'idle', value: null, error: null, selectedId: null, requestId: 0 },
 }
 
-function requestPhase<T>(slot: ReadSlot<T>, silent: boolean | undefined): ReadSlot<T> {
+function requestPhase<S extends ReadSlot<unknown>>(slot: S, silent: boolean | undefined): S {
   return { ...slot, phase: silent && slot.phase === 'ready' ? 'refreshing' : 'loading', error: null }
 }
 
@@ -52,7 +59,7 @@ export function intelligentEvalReadReducer(
     case 'list_requested':
       return { ...state, list: requestPhase(state.list, action.silent) }
     case 'list_succeeded':
-      return { ...state, list: { phase: 'ready', value: action.value, error: null } }
+      return { ...state, list: { phase: 'ready', value: action.value, error: null, total: action.total } }
     case 'list_failed':
       return {
         ...state,

frontend/web/src/read/useIntelligentEvalRead.test.tsx

@@ -28,19 +28,20 @@ describe('useIntelligentEvalRead', () => {
   it('refreshes active evaluations every five seconds and stops at terminal state', async () => {
     vi.useFakeTimers()
     const adapter: IntelligentEvalReadAdapter = {
-      list: vi.fn().mockResolvedValue([evaluation]),
+      list: vi.fn().mockResolvedValue({ items: [evaluation], total: 1 }),
       get: vi.fn().mockResolvedValue(evaluation),
     }
-    renderHook(() => useIntelligentEvalRead(null, adapter))
+    const { result } = renderHook(() => useIntelligentEvalRead(null, adapter))
     await settle()
     expect(adapter.list).toHaveBeenCalledTimes(1)
+    expect(result.current.list.total).toBe(1)
 
     await act(async () => { await vi.advanceTimersByTimeAsync(5000) })
     await settle()
     expect(adapter.list).toHaveBeenCalledTimes(2)
 
     const terminal = { ...evaluation, status: 'completed' } as IntelligentEval
-    vi.mocked(adapter.list).mockResolvedValue([terminal])
+    vi.mocked(adapter.list).mockResolvedValue({ items: [terminal], total: 1 })
     await act(async () => { await vi.advanceTimersByTimeAsync(5000) })
     await settle()
     await act(async () => { await vi.advanceTimersByTimeAsync(10000) })
@@ -51,7 +52,7 @@ describe('useIntelligentEvalRead', () => {
     vi.useFakeTimers()
     const adapter: IntelligentEvalReadAdapter = {
       list: vi.fn()
-        .mockResolvedValueOnce([evaluation])
+        .mockResolvedValueOnce({ items: [evaluation], total: 1 })
         .mockRejectedValueOnce(new Error('网络错误')),
       get: vi.fn().mockResolvedValue(evaluation),
     }
@@ -65,12 +66,29 @@ describe('useIntelligentEvalRead', () => {
     expect(result.current.list.error).toBeNull()
   })
 
+  it('passes page/pageSize to the adapter and reloads on change', async () => {
+    const adapter: IntelligentEvalReadAdapter = {
+      list: vi.fn().mockResolvedValue({ items: [], total: 0 }),
+      get: vi.fn().mockResolvedValue(evaluation),
+    }
+    const { rerender } = renderHook(
+      ({ page }) => useIntelligentEvalRead(null, adapter, page, 20),
+      { initialProps: { page: 1 } },
+    )
+    await settle()
+    expect(adapter.list).toHaveBeenCalledWith(1, 20)
+
+    rerender({ page: 2 })
+    await settle()
+    expect(adapter.list).toHaveBeenCalledWith(2, 20)
+  })
+
   it('clears the old detail and ignores a response for the previous selection', async () => {
     const first = deferred<IntelligentEval>()
     const second = deferred<IntelligentEval>()
     const secondEvaluation = { ...evaluation, id: 'eval-2', name: '评估二' } as IntelligentEval
     const adapter: IntelligentEvalReadAdapter = {
-      list: vi.fn().mockResolvedValue([]),
+      list: vi.fn().mockResolvedValue({ items: [], total: 0 }),
       get: vi.fn((id: string) => (id === 'eval-1' ? first.promise : second.promise)),
     }
     const { result, rerender } = renderHook(
@@ -91,7 +109,7 @@ describe('useIntelligentEvalRead', () => {
 
   it('surfaces failure for a newly selected detail instead of retaining the old snapshot', async () => {
     const adapter: IntelligentEvalReadAdapter = {
-      list: vi.fn().mockResolvedValue([]),
+      list: vi.fn().mockResolvedValue({ items: [], total: 0 }),
       get: vi.fn()
         .mockResolvedValueOnce(evaluation)
         .mockRejectedValueOnce(new Error('不存在')),

frontend/web/src/read/useIntelligentEvalRead.ts

@@ -21,6 +21,8 @@ function isActive(status: string | undefined): boolean {
 export function useIntelligentEvalRead(
   selectedId: string | null,
   adapter: IntelligentEvalReadAdapter = intelligentEvalReadAdapter,
+  page: number = 1,
+  pageSize: number = 20,
 ): IntelligentEvalReadState & { reloadList: () => Promise<void>; reloadDetail: () => Promise<void> } {
   const [state, dispatch] = useReducer(intelligentEvalReadReducer, initialIntelligentEvalReadState)
   const detailRequestId = useRef(0)
@@ -28,12 +30,12 @@ export function useIntelligentEvalRead(
   const loadList = useCallback(async (silent = false) => {
     dispatch({ type: 'list_requested', silent })
     try {
-      const value = await adapter.list()
-      dispatch({ type: 'list_succeeded', value })
+      const { items, total } = await adapter.list(page, pageSize)
+      dispatch({ type: 'list_succeeded', value: items, total })
     } catch (error) {
       dispatch({ type: 'list_failed', error: errorMessage(error) })
     }
-  }, [adapter])
+  }, [adapter, page, pageSize])
 
   const loadDetail = useCallback(async (id: string, silent = false) => {
     const requestId = ++detailRequestId.current

tests/integration/test_intelligent_eval_task_queue_api.py

@@ -274,3 +274,63 @@ def test_list_tasks_not_shadowed_by_eval_id(client: TestClient):
     response = client.get("/api/intelligent-evals/tasks")
     assert response.status_code == 200
     assert "tasks" in response.json()
+
+
+def _make_eval(db_session: Session, name: str, days_ago: int) -> IntelligentEvalDB:
+    from agenteval.storage.db import utc_now
+
+    ev = IntelligentEvalDB(
+        name=name,
+        target_id="target1",
+        status=IntelligentEvalStatus.COMPLETED.value,
+        created_at=utc_now() - timedelta(days=days_ago),
+        started_at=utc_now() - timedelta(days=days_ago),
+    )
+    db_session.add(ev)
+    return ev
+
+
+def test_list_evals_backward_compatible_no_page(client: TestClient, db_session: Session):
+    """Without page param the endpoint returns everything (no total key)."""
+    _make_eval(db_session, "A", days_ago=3)
+    _make_eval(db_session, "B", days_ago=2)
+    db_session.commit()
+
+    response = client.get("/api/intelligent-evals")
+    assert response.status_code == 200
+    data = response.json()
+    assert len(data["intelligent_evals"]) == 2
+    assert "total" not in data
+
+
+def test_list_evals_pagination(client: TestClient, db_session: Session):
+    """page/page_size returns one slice (newest first) plus total."""
+    _make_eval(db_session, "oldest", days_ago=3)
+    _make_eval(db_session, "middle", days_ago=2)
+    _make_eval(db_session, "newest", days_ago=1)
+    db_session.commit()
+
+    response = client.get("/api/intelligent-evals?page=1&page_size=2")
+    assert response.status_code == 200
+    data = response.json()
+    assert data["total"] == 3
+    assert data["page"] == 1
+    assert data["page_size"] == 2
+    names = [e["name"] for e in data["intelligent_evals"]]
+    assert names == ["newest", "middle"]  # created_at desc
+
+    response = client.get("/api/intelligent-evals?page=2&page_size=2")
+    names = [e["name"] for e in response.json()["intelligent_evals"]]
+    assert names == ["oldest"]
+
+
+def test_list_evals_page_size_clamped(client: TestClient, db_session: Session):
+    """page_size is clamped to [1, 100]."""
+    for i in range(5):
+        _make_eval(db_session, f"eval{i}", days_ago=10 - i)
+    db_session.commit()
+
+    response = client.get("/api/intelligent-evals?page=1&page_size=0")
+    data = response.json()
+    assert data["page_size"] == 1
+    assert len(data["intelligent_evals"]) == 1