diff --git a/tests/unit/test_intelligent_eval_scheduler.py b/tests/unit/test_intelligent_eval_scheduler.py
index ff7e699..9af50a5 100644
--- a/tests/unit/test_intelligent_eval_scheduler.py
+++ b/tests/unit/test_intelligent_eval_scheduler.py
@@ -329,3 +329,187 @@ def test_fire_and_forget_does_not_block_and_logs_errors(monkeypatch):
         logger.removeHandler(handler)
         logger.disabled = was_disabled
     assert any("触发爆炸" in r.getMessage() for r in records)
+
+
+def test_fire_and_forget_cancelled_task_does_not_log(monkeypatch):
+    """_fire_and_forget: cancelled task 直接 return，不记录异常。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    async def long_running():
+        await asyncio.sleep(100)
+
+    async def main():
+        records: list[logging.LogRecord] = []
+        handler = logging.Handler()
+        handler.emit = records.append
+        logger = logging.getLogger("agenteval")
+        was_disabled = logger.disabled
+        logger.disabled = False
+        logger.addHandler(handler)
+        try:
+            scheduler_mod._fire_and_forget(long_running(), "Worker")
+            await asyncio.sleep(0.01)  # 让 task 启动
+            # 取消所有非守护 task
+            tasks = [t for t in asyncio.all_tasks() if t is not asyncio.current_task()]
+            for t in tasks:
+                t.cancel()
+            await asyncio.gather(*tasks, return_exceptions=True)
+            await asyncio.sleep(0.01)  # 让完成回调执行
+        finally:
+            logger.removeHandler(handler)
+            logger.disabled = was_disabled
+        # cancelled task 不应记录异常
+        assert not any("触发任务异常" in r.getMessage() for r in records)
+
+    asyncio.run(main())
+
+
+def test_trigger_openclaw_agent_nonzero_exit_calls_on_failure(monkeypatch):
+    """_trigger_openclaw_agent: 子进程非零退出码 → on_failure 回调。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    def fake_run(cmd, **kwargs):
+        return MagicMock(returncode=1, stderr="docker: command not found")
+
+    monkeypatch.setattr(subprocess, "run", fake_run)
+
+    failures: list[str] = []
+    asyncio.run(scheduler_mod._trigger_openclaw_agent("Worker", "test", "msg", failures.append))
+
+    assert len(failures) == 1
+    assert "exit=1" in failures[0]
+    assert "docker: command not found" in failures[0]
+
+
+def test_trigger_openclaw_agent_exception_calls_on_failure(monkeypatch):
+    """_trigger_openclaw_agent: subprocess.run 抛异常 → on_failure 回调。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    def fake_run(cmd, **kwargs):
+        raise FileNotFoundError("docker not found")
+
+    monkeypatch.setattr(subprocess, "run", fake_run)
+
+    failures: list[str] = []
+    asyncio.run(scheduler_mod._trigger_openclaw_agent("Worker", "test", "msg", failures.append))
+
+    assert len(failures) == 1
+    assert "docker not found" in failures[0]
+
+
+def test_record_worker_trigger_failure_records_to_pending_evals(monkeypatch):
+    """_record_worker_trigger_failure: 有待认领任务的评估 → 落账触发失败。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    mock_session = MagicMock()
+    monkeypatch.setattr(scheduler_mod, "get_session", lambda: mock_session)
+    monkeypatch.setattr(scheduler_mod, "eval_ids_with_pending_worker_tasks", lambda session: ["ev-1", "ev-2"])
+
+    recorded: list = []
+    monkeypatch.setattr(
+        scheduler_mod,
+        "record_trigger_failures",
+        lambda session, channel, eval_ids, error: recorded.append((channel, eval_ids, error)),
+    )
+
+    scheduler_mod._record_worker_trigger_failure("test error")
+
+    assert len(recorded) == 1
+    assert recorded[0] == ("worker", ["ev-1", "ev-2"], "test error")
+    mock_session.close.assert_called_once()
+
+
+def test_record_worker_trigger_failure_skips_when_no_pending(monkeypatch):
+    """_record_worker_trigger_failure: 无待认领任务 → 不落账。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    mock_session = MagicMock()
+    monkeypatch.setattr(scheduler_mod, "get_session", lambda: mock_session)
+    monkeypatch.setattr(scheduler_mod, "eval_ids_with_pending_worker_tasks", lambda session: [])
+
+    recorded: list = []
+    monkeypatch.setattr(
+        scheduler_mod,
+        "record_trigger_failures",
+        lambda session, channel, eval_ids, error: recorded.append((channel, eval_ids, error)),
+    )
+
+    scheduler_mod._record_worker_trigger_failure("test error")
+
+    assert len(recorded) == 0
+    mock_session.close.assert_called_once()
+
+
+def test_record_worker_trigger_failure_swallows_db_errors(monkeypatch):
+    """_record_worker_trigger_failure: DB 异常不传播（兜底日志）。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    mock_session = MagicMock()
+    monkeypatch.setattr(scheduler_mod, "get_session", lambda: mock_session)
+    monkeypatch.setattr(
+        scheduler_mod,
+        "eval_ids_with_pending_worker_tasks",
+        lambda session: (_ for _ in ()).throw(RuntimeError("DB error")),
+    )
+
+    # 不应抛异常
+    scheduler_mod._record_worker_trigger_failure("test error")
+    mock_session.close.assert_called_once()
+
+
+def test_record_planner_trigger_failure_records_to_planning_evals(monkeypatch):
+    """_record_planner_trigger_failure: 有 planning 评估 → 落账触发失败。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    mock_session = MagicMock()
+    mock_row = MagicMock()
+    mock_row.id = "ev-plan-1"
+    mock_session.exec.return_value.all.return_value = [mock_row]
+    monkeypatch.setattr(scheduler_mod, "get_session", lambda: mock_session)
+
+    recorded: list = []
+    monkeypatch.setattr(
+        scheduler_mod,
+        "record_trigger_failures",
+        lambda session, channel, eval_ids, error: recorded.append((channel, eval_ids, error)),
+    )
+
+    scheduler_mod._record_planner_trigger_failure("planner error")
+
+    assert len(recorded) == 1
+    assert recorded[0] == ("planner", ["ev-plan-1"], "planner error")
+    mock_session.close.assert_called_once()
+
+
+def test_record_planner_trigger_failure_skips_when_no_planning(monkeypatch):
+    """_record_planner_trigger_failure: 无 planning 评估 → 不落账。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    mock_session = MagicMock()
+    mock_session.exec.return_value.all.return_value = []
+    monkeypatch.setattr(scheduler_mod, "get_session", lambda: mock_session)
+
+    recorded: list = []
+    monkeypatch.setattr(
+        scheduler_mod,
+        "record_trigger_failures",
+        lambda session, channel, eval_ids, error: recorded.append((channel, eval_ids, error)),
+    )
+
+    scheduler_mod._record_planner_trigger_failure("planner error")
+
+    assert len(recorded) == 0
+    mock_session.close.assert_called_once()
+
+
+def test_record_planner_trigger_failure_swallows_db_errors(monkeypatch):
+    """_record_planner_trigger_failure: DB 异常不传播（兜底日志）。"""
+    import agenteval.intelligent_eval.scheduler as scheduler_mod
+
+    mock_session = MagicMock()
+    mock_session.exec.side_effect = RuntimeError("DB error")
+    monkeypatch.setattr(scheduler_mod, "get_session", lambda: mock_session)
+
+    # 不应抛异常
+    scheduler_mod._record_planner_trigger_failure("planner error")
+    mock_session.close.assert_called_once()