diff --git a/debug/accuracy_tools/msprobe/pytorch/monitor/anomaly_analyse.py b/debug/accuracy_tools/msprobe/pytorch/monitor/anomaly_analyse.py
index fc21ded9515ed1c792c72d3de423e8dcf8e56e7c..6fec808d14072b46b15fbdba8541727413d6a530 100644
--- a/debug/accuracy_tools/msprobe/pytorch/monitor/anomaly_analyse.py
+++ b/debug/accuracy_tools/msprobe/pytorch/monitor/anomaly_analyse.py
@@ -65,6 +65,9 @@ class AnomalyDataWriter:
         Args:
         anomalies: GradAnomalyData对象列表
         """
+        if not isinstance(anomalies, list) or anomalies == []:
+            logger.warning("The anomalies should be a list of GradAnomalyData and not empty.")
+            return
         anomalies_json = self.get_anomaly_dict(anomalies)
         logger.info(f"{MonitorConst.ANOMALY_JSON} is at {self.dump_rank_dir}.")
 
diff --git a/debug/accuracy_tools/msprobe/pytorch/monitor/module_hook.py b/debug/accuracy_tools/msprobe/pytorch/monitor/module_hook.py
index cb40b99be027a96c77630e47d7384af4b8045c1b..ab80307e344afdad9cf06b6a7822ace654be9cff 100644
--- a/debug/accuracy_tools/msprobe/pytorch/monitor/module_hook.py
+++ b/debug/accuracy_tools/msprobe/pytorch/monitor/module_hook.py
@@ -228,6 +228,8 @@ class TrainerMon:
                 self.anomaly_data_factory,
                 self.ndigits
             )
+            # 初始化anomaly_detect状态
+            self.is_anomaly_detected = False
             # 初始化anomaly detected文件目录
             if self.anomaly_data_factory:
                 self.anomaly_data_writer = AnomalyDataWriter(os.path.join(output_base_dir, "anomaly_detected"), rank)
@@ -491,6 +493,9 @@ class TrainerMon:
             self.write_metrics(self.ops, self.summary_writer, self.grad_context.acc_metric, step, 'grad_unreduced')
         self.write_metrics(self.ops, self.summary_writer, self.grad_context.post, step, 'grad_reduced')
 
+    def get_anomaly_status(self):
+        return self.is_anomaly_detected
+
     def hook_optimizer(self, optimizer=None):
         # in DDP by default use params_have_main_grad
         def optimizer_pre_step_hook(optimizer, args, kwargs):
@@ -547,6 +552,8 @@ class TrainerMon:
                 metric_dict.update(cc.data)
                 cc.reset()
 
+            # clear anomaly detection status
+            self.is_anomaly_detected = False
             if not metric_dict:
                 return
             context.metric_dict = metric_dict
@@ -576,8 +583,13 @@ class TrainerMon:
                 self.write_metrics(self.ops, self.summary_writer, context.metric_dict, context.step, 'other')
             context.metric_dict.clear()
             context.step += 1
+
+            anomaly_data = self.summary_writer.get_anomalies()
+            if anomaly_data:
+                # set anomaly detected status
+                self.is_anomaly_detected = True
             if self.anomaly_data_factory:
-                self.anomaly_data_writer.write_detected_json(self.summary_writer.get_anomalies())
+                self.anomaly_data_writer.write_detected_json(anomaly_data)
             self.summary_writer.clear_anomalies()
             self.call_id = 0
             return