fix(client): resolve trajectory review follow-ups

packages/client/runtime/tests/fold-adapter.spec.ts

@@ -178,6 +178,45 @@ describe('FoldAdapter', () => {
     expect(adapter.nodes().nodes[0]).toMatchObject({ kind: 'tool-result', isError: true, error: { code: 'boom' } })
   })
 
+  it('indexes assistant timing and the active request header in one replay pass', () => {
+    const adapter = new FoldAdapter()
+    adapter.reset([
+      ev.stepStart(0, 1, 2),
+      at(1, { type: 'request/header', data: {
+        reason: 'initial',
+        header: {
+          config: { provider: 'fake', model: 'first' },
+          tools: [],
+        },
+      } }),
+      ev.chunkStart(2, 1, 2),
+      ev.chunkText(3, 1, 'token', 2),
+      ev.assistant(4, 1, 'done', 2),
+    ], 0)
+
+    expect(adapter.nodes().nodes[0]).toMatchObject({
+      kind: 'assistant',
+      timing: {
+        stepStartTime: 1_700_000_000_000,
+        firstTokenTime: 1_700_000_000_003,
+        completedTime: 1_700_000_000_004,
+      },
+      requestConfig: { provider: 'fake', model: 'first' },
+    })
+
+    adapter.append(ev.stepStart(5, 2, 1))
+    adapter.append(ev.chunkText(6, 2, 'next', 1))
+    adapter.append(ev.assistant(7, 2, 'next done', 1))
+    expect(adapter.nodes().nodes.at(-1)).toMatchObject({
+      timing: {
+        stepStartTime: 1_700_000_000_005,
+        firstTokenTime: 1_700_000_000_006,
+        completedTime: 1_700_000_000_007,
+      },
+      requestConfig: { provider: 'fake', model: 'first' },
+    })
+  })
+
   it('exposes the in-window call index for runningCalls material', () => {
     const adapter = new FoldAdapter()
     adapter.reset([ev.toolCall(0, 1, 'c9', 'slow', '{}')], 0)

packages/client/runtime/tests/request-inspection.spec.ts

@@ -109,6 +109,56 @@ describe('inspectRequests', () => {
     expect(snapshot.callSchemas.get('nested')?.name).toBe('read')
   })
 
+  it('keeps chunk-reported usage through request failure and prefers it to message fallback', () => {
+    const chunkUsage = { inputTokens: 21, outputTokens: 3 }
+    const retryUsage = {
+      inputTokens: 5,
+      outputTokens: 2,
+      cacheReadTokens: 8,
+      reasoningTokens: 1,
+    }
+    const snapshot = inspectRequests(entriesOf([
+      at(0, 'step/start', { turn: 1, step: 1 }),
+      at(1, 'assistant/chunk', {
+        turn: 1,
+        step: 1,
+        chunk: { type: 'usage', usage: chunkUsage },
+      }),
+      at(2, 'llm/retry', {
+        turn: 1,
+        step: 1,
+        retry: 1,
+        maxRetries: 2,
+        delayMs: 100,
+        failure: { message: 'rate limited' },
+      }),
+      at(3, 'assistant/chunk', {
+        turn: 1,
+        step: 1,
+        chunk: { type: 'usage', usage: retryUsage },
+      }),
+      at(4, 'assistant/message', {
+        turn: 1,
+        step: 1,
+        message: createAssistantMessage({
+          content: [{ type: 'text', text: 'recovered' }],
+          source: { provider: 'fake', model: 'model' },
+        }),
+        usage: { inputTokens: 1, outputTokens: 1 },
+      }),
+    ]))
+
+    expect(snapshot.requests[0]).toMatchObject({
+      status: 'complete',
+      usage: {
+        inputTokens: 26,
+        outputTokens: 5,
+        cacheReadTokens: 8,
+        reasoningTokens: 1,
+      },
+    })
+  })
+
   it('treats a scrubbed durable-fixture tool catalog as unavailable', () => {
     const snapshot = inspectRequests(entriesOf([
       at(0, 'step/start', { turn: 1, step: 1 }),

packages/client/runtime/tests/session.spec.ts

@@ -292,6 +292,26 @@ describe('paging', () => {
       .toEqual([1, 3, 7, 9, 13, 15])
   })
 
+  it('stops complete-history loading between pages after its consumer aborts', async () => {
+    const middle = deferred<Awaited<ReturnType<FakeApiClient['onHistory']>>>()
+    const { api, session } = makeSession()
+    api.onHistory = payload => payload.beforeSeq === undefined
+      ? histResponse(plainTurn(12, 2, '最新问', '最新答'), true)
+      : middle.promise
+    await session.open()
+    const controller = new AbortController()
+    const completeHistory = session.loadAllHistory(controller.signal)
+    controller.abort()
+    middle.resolve(ok({
+      events: entries(plainTurn(6, 1, '中间问', '中间答')) as never[],
+      hasMore: true,
+    }))
+    await completeHistory
+
+    expect(api.callsOf('session.history')).toHaveLength(2)
+    expect(session.getSnapshot().hasMore).toBe(true)
+  })
+
   it('drops a discontinuous older page fail-soft (window unchanged, hasMore cleared)', async () => {
     const { api, session } = makeSession()
     api.onHistory = payload => payload.beforeSeq === undefined
@@ -767,6 +787,32 @@ describe('resync', () => {
     expect(snapshot.openState).toBe('open') // stale failure did not settle the fresh generation into error
     expect(snapshot.nodes.map(n => n.seq)).toEqual([7, 9])
   })
+
+  it('starts fresh paging while an older generation page is still pending', async () => {
+    const stalePage = deferred<Awaited<ReturnType<FakeApiClient['onHistory']>>>()
+    const { api, session } = makeSession()
+    let call = 0
+    api.onHistory = () => {
+      call++
+      if (call === 1) return histResponse(plainTurn(6, 1, '新', '页'), true)
+      if (call === 2) return stalePage.promise
+      if (call === 3) return histResponse(plainTurn(6, 1, '新', '代'), true)
+      return histResponse(plainTurn(0, 0, '旧', '页'), false)
+    }
+    await session.open()
+    const stale = session.loadOlder()
+    await session.resync()
+    const fresh = session.loadAllHistory()
+    await vi.waitFor(() => { expect(call).toBe(4) })
+    stalePage.resolve(ok({
+      events: entries(plainTurn(0, 0, '废', '弃')) as never[],
+      hasMore: false,
+    }))
+    await Promise.all([stale, fresh])
+
+    expect(session.getSnapshot().hasMore).toBe(false)
+    expect(session.getSnapshot().nodes.map(node => node.seq)).toEqual([1, 3, 7, 9])
+  })
 })
 
 describe('run_code sub-dispatch indexing', () => {