diff --git a/api/app/controllers/user_memory_controllers.py b/api/app/controllers/user_memory_controllers.py
index c54ebbd1..da12cbf6 100644
--- a/api/app/controllers/user_memory_controllers.py
+++ b/api/app/controllers/user_memory_controllers.py
@@ -11,6 +11,7 @@ from app.db import get_db
 from app.core.logging_config import get_api_logger
 from app.core.response_utils import success, fail
 from app.core.error_codes import BizCode
+from app.core.api_key_utils import timestamp_to_datetime
 from app.services.user_memory_service import (
     UserMemoryService,
     analytics_memory_types,
@@ -356,7 +357,7 @@ async def update_end_user_profile(
         if 'hire_date' in update_data:
             hire_date_timestamp = update_data['hire_date']
             if hire_date_timestamp is not None:
-                update_data['hire_date'] = UserMemoryService.timestamp_to_datetime(hire_date_timestamp)
+                update_data['hire_date'] = timestamp_to_datetime(hire_date_timestamp)
             # 如果是 None，保持 None（允许清空）
         
         for field, value in update_data.items():
diff --git a/api/app/core/memory/utils/prompt/prompts/user_summary.jinja2 b/api/app/core/memory/utils/prompt/prompts/user_summary.jinja2
index 373ab31e..2f452c53 100644
--- a/api/app/core/memory/utils/prompt/prompts/user_summary.jinja2
+++ b/api/app/core/memory/utils/prompt/prompts/user_summary.jinja2
@@ -85,33 +85,21 @@ Example Output:
 ===End of Example===
 
 
-===Reflection Process===
+===Internal Quality Checks (DO NOT OUTPUT)===
 
-After generating the profile, perform the following self-review steps:
+Before generating your final output, internally verify:
+1. All content is grounded in provided data (no fabrication)
+2. Format follows the specified structure with correct headers
+3. Tone is objective, third-person, and neutral
+4. All four sections are complete and within character limits
 
-**Step 1: Data Grounding Check**
-- Verify all statements are supported by the provided entities and statements
-- Ensure no fabricated or speculated information is included
-- Confirm all claims can be traced back to the input data
-
-**Step 2: Format Compliance**
-- Verify each section follows the specified format with section headers
-- Check character count limits for each section
-- Ensure proper use of section markers (【】)
-
-**Step 3: Tone and Style Review**
-- Confirm objective third-person perspective is maintained
-- Check for excessive adjectives or empty phrases
-- Verify neutral and restrained tone throughout
-
-**Step 4: Completeness Check**
-- Ensure all four sections are present and complete
-- Verify each section addresses its specific focus area
-- Confirm the one-sentence summary effectively captures the user's essence
+**IMPORTANT: These checks are for your internal use only. DO NOT include them in your output.**
 
 
 ===Output Requirements===
 
+**CRITICAL: Your response must ONLY contain the four sections below. Do not include any reflection, self-review, or meta-commentary.**
+
 **LANGUAGE REQUIREMENT:**
 - The output language should ALWAYS be Chinese (Simplified)
 - All section content must be in Chinese
@@ -122,3 +110,5 @@ After generating the profile, perform the following self-review steps:
 - Content follows immediately after the header
 - Sections are separated by blank lines
 - Strictly adhere to character limits for each section
+- **DO NOT include any text after the 【一句话总结】 section**
+- **DO NOT output reflection steps, self-review, or verification notes**
diff --git a/api/app/services/user_memory_service.py b/api/app/services/user_memory_service.py
index 6fd72fce..40851835 100644
--- a/api/app/services/user_memory_service.py
+++ b/api/app/services/user_memory_service.py
@@ -1054,6 +1054,28 @@ async def analytics_user_summary(end_user_id: Optional[str] = None) -> Dict[str,
         core_values = core_values_match.group(1).strip() if core_values_match else ""
         one_sentence = one_sentence_match.group(1).strip() if one_sentence_match else ""
         
+        # 6) 清理可能包含的反思内容（防御性编程）
+        # 如果 LLM 仍然输出了反思内容，在这里过滤掉
+        def clean_reflection_content(text: str) -> str:
+            """移除可能包含的反思内容"""
+            if not text:
+                return text
+            # 移除 "---" 之后的所有内容（通常是反思部分的开始）
+            if '---' in text:
+                text = text.split('---')[0].strip()
+            # 移除 "**Step" 开头的内容
+            if '**Step' in text:
+                text = text.split('**Step')[0].strip()
+            # 移除 "Self-Review" 相关内容
+            if 'Self-Review' in text or 'self-review' in text:
+                text = re.sub(r'[\-\*]*\s*Self-Review.*$', '', text, flags=re.IGNORECASE | re.DOTALL).strip()
+            return text
+        
+        user_summary = clean_reflection_content(user_summary)
+        personality = clean_reflection_content(personality)
+        core_values = clean_reflection_content(core_values)
+        one_sentence = clean_reflection_content(one_sentence)
+        
         return {
             "user_summary": user_summary,
             "personality": personality,