From 8ea243c572c65746acc3422d7e4c6f8d0344738d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=9D=8E=E6=96=B0=E6=9C=88?=
 <accounts_68c0f5d519f260d93ee2997e@mail.teambition.com>
Date: Tue, 30 Dec 2025 08:37:37 +0000
Subject: [PATCH 1/8] Merge #87 into develop from fix/develop_kj_knowledge
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Agent应用中添加知识库的配置字段（提示词修改、反思给默认值）

* fix/develop_kj_knowledge: (1 commits squashed)

  - Agent应用中添加知识库的配置字段（提示词修改、反思给默认值）

Signed-off-by: aliyun8644380055 <accounts_68c0f5d519f260d93ee2997e@mail.teambition.com>
Reviewed-by: aliyun6762716068 <accounts_68cb7c6b61f5dcc4200d6251@mail.teambition.com>
Merged-by: aliyun6762716068 <accounts_68cb7c6b61f5dcc4200d6251@mail.teambition.com>

CR-link: https://codeup.aliyun.com/redbearai/python/redbear-mem-open/change/87
---
 api/app/core/memory/utils/prompt/prompts/evaluate.jinja2 | 2 +-
 api/app/schemas/app_schema.py                            | 1 +
 api/app/schemas/memory_reflection_schemas.py             | 4 ++--
 3 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/api/app/core/memory/utils/prompt/prompts/evaluate.jinja2 b/api/app/core/memory/utils/prompt/prompts/evaluate.jinja2
index 200f2667..e649897a 100644
--- a/api/app/core/memory/utils/prompt/prompts/evaluate.jinja2
+++ b/api/app/core/memory/utils/prompt/prompts/evaluate.jinja2
@@ -86,5 +86,5 @@
 - **quality_assessment**:
     quality_assessment=true时输出评估对象，否则为null（注意：- summary输出的结果不允许含有（expired_at设为2024-01-01T00:00:00Z)等原数据字段以及涉及需要修改的字段以及内容）
 - **memory_verify**: memory_verify=true时输出隐私检测对象，否则为null
-    （注意：- summary输出的结果不允许含有（expired_at设为2024-01-01T00:00:00Z)等原数据字段以及涉及需要修改的字段以及内容）
+    （注意：- summary输出的结果不允许含有（expired_at设为2024-01-01T00:00:00Z、memory_verify=true\memory_verify=false)等原数据字段以及涉及需要修改的字段以及内容）
 模式参考：{{ json_schema }}
\ No newline at end of file
diff --git a/api/app/schemas/app_schema.py b/api/app/schemas/app_schema.py
index de0a4c53..81cd704d 100644
--- a/api/app/schemas/app_schema.py
+++ b/api/app/schemas/app_schema.py
@@ -32,6 +32,7 @@ class KnowledgeRetrievalConfig(BaseModel):
     )
     reranker_id: Optional[str] = Field(default=None, description="多知识库结果融合的模型ID")
     reranker_top_k: int = Field(default=10, ge=0, le=1024, description="多知识库结果融合的模型参数")
+    use_graph: bool = Field(default=False, description="是否使用图搜索")
 
 
 class ToolConfig(BaseModel):
diff --git a/api/app/schemas/memory_reflection_schemas.py b/api/app/schemas/memory_reflection_schemas.py
index ada92cf2..860f1ef1 100644
--- a/api/app/schemas/memory_reflection_schemas.py
+++ b/api/app/schemas/memory_reflection_schemas.py
@@ -12,8 +12,8 @@ class Memory_Reflection(BaseModel):
     config_id: Optional[int] = None
     reflection_enabled: bool
     reflection_period_in_hours: str
-    reflexion_range: str
-    baseline: str
+    reflexion_range: Optional[str] = "partial"
+    baseline: Optional[str] = "TIME"
     reflection_model_id: str
     memory_verify: bool
     quality_assessment: bool

From 82b9925448f684f0768d7e7072d13eaa3f1fcd2a Mon Sep 17 00:00:00 2001
From: lixiangcheng1 <lixiangcheng1@wanda.cn>
Date: Tue, 30 Dec 2025 17:44:37 +0800
Subject: [PATCH 2/8] [fix]document:pandas.read_excel error: Missing optional
 dependency 'python-calamine'.

---
 api/app/models/document_model.py  | 5 ++++-
 api/app/models/knowledge_model.py | 5 ++++-
 api/pyproject.toml                | 2 ++
 api/requirements.txt              | 2 ++
 4 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/api/app/models/document_model.py b/api/app/models/document_model.py
index db9280c6..fb43d44d 100644
--- a/api/app/models/document_model.py
+++ b/api/app/models/document_model.py
@@ -26,6 +26,7 @@ class Document(Base):
                                "html4excel": False,
                                "graphrag": {
                                     "use_graphrag": False,
+                                    "scene_name": "",
                                     "entity_types": [
                                         "organization",
                                         "person",
@@ -33,7 +34,9 @@ class Document(Base):
                                         "event",
                                         "category"
                                     ],
-                                    "method": "general"
+                                    "method": "general",
+                                    "resolution": True,
+                                    "community": True
                                 }
                            }, comment="default parser config")
     chunk_num = Column(Integer, default=0, comment="chunk num")
diff --git a/api/app/models/knowledge_model.py b/api/app/models/knowledge_model.py
index 6d3465f9..8f0909d3 100644
--- a/api/app/models/knowledge_model.py
+++ b/api/app/models/knowledge_model.py
@@ -65,6 +65,7 @@ class Knowledge(Base):
                                "html4excel": False,
                                "graphrag": {
                                     "use_graphrag": False,
+                                    "scene_name": "",
                                     "entity_types": [
                                         "organization",
                                         "person",
@@ -72,7 +73,9 @@ class Knowledge(Base):
                                         "event",
                                         "category"
                                     ],
-                                    "method": "general"
+                                    "method": "general",
+                                    "resolution": True,
+                                    "community": True
                                 }
                            },
                            comment="default parser config")
diff --git a/api/pyproject.toml b/api/pyproject.toml
index 901858e6..2dcc706d 100644
--- a/api/pyproject.toml
+++ b/api/pyproject.toml
@@ -135,6 +135,8 @@ dependencies = [
     "graspologic==3.4.5.dev2",
     "markdown-to-json==2.1.1",
     "valkey==6.0.2",
+    "python-calamine>=0.4.0",
+    "xlrd==2.0.2"
 ]
 
 [tool.pytest.ini_options]
diff --git a/api/requirements.txt b/api/requirements.txt
index 5530a9e3..99252e09 100644
--- a/api/requirements.txt
+++ b/api/requirements.txt
@@ -129,3 +129,5 @@ editdistance==0.8.1
 graspologic==3.4.5.dev2
 markdown-to-json==2.1.1
 valkey==6.0.2
+python-calamine>=0.4.0
+xlrd==2.0.2

From 909c536b473f554d1bab52fea7b9685bbaa950ca Mon Sep 17 00:00:00 2001
From: lixiangcheng1 <lixiangcheng1@wanda.cn>
Date: Tue, 30 Dec 2025 18:20:34 +0800
Subject: [PATCH 3/8] [fix]parsed excel document error:float division by zero

---
 api/app/core/rag/deepdoc/parser/excel_parser.py | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/api/app/core/rag/deepdoc/parser/excel_parser.py b/api/app/core/rag/deepdoc/parser/excel_parser.py
index b6e1e4a1..856155f1 100644
--- a/api/app/core/rag/deepdoc/parser/excel_parser.py
+++ b/api/app/core/rag/deepdoc/parser/excel_parser.py
@@ -52,13 +52,19 @@ class RAGExcelParser:
                 raise Exception(f"pandas.read_excel error: {e_pandas}, original openpyxl error: {e}")
 
     @staticmethod
-    def _clean_dataframe(df: pd.DataFrame):
+    def _clean_dataframe(df):
         def clean_string(s):
             if isinstance(s, str):
                 return ILLEGAL_CHARACTERS_RE.sub(" ", s)
             return s
 
-        return df.apply(lambda col: col.map(clean_string))
+        # 处理单 DataFrame 或字典（多 Sheet）
+        if isinstance(df, dict):
+            return {sheet: RAGExcelParser._clean_dataframe(sheet_df) for sheet, sheet_df in df.items()}
+        elif isinstance(df, pd.DataFrame):
+            return df.apply(lambda col: col.map(clean_string))
+        else:
+            raise ValueError(f"Unsupported type for cleaning: {type(df)}")
 
     @staticmethod
     def _dataframe_to_workbook(df):

From 775d36b16b6fa0566ebd50bb8d1284752c4625d1 Mon Sep 17 00:00:00 2001
From: lixiangcheng1 <lixiangcheng1@wanda.cn>
Date: Tue, 30 Dec 2025 19:05:29 +0800
Subject: [PATCH 4/8] [fix]parsed excel document error:float division by zero

---
 api/app/core/rag/deepdoc/parser/excel_parser.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/api/app/core/rag/deepdoc/parser/excel_parser.py b/api/app/core/rag/deepdoc/parser/excel_parser.py
index 856155f1..07e39f4a 100644
--- a/api/app/core/rag/deepdoc/parser/excel_parser.py
+++ b/api/app/core/rag/deepdoc/parser/excel_parser.py
@@ -69,7 +69,8 @@ class RAGExcelParser:
     @staticmethod
     def _dataframe_to_workbook(df):
         # if contains multiple sheets use _dataframes_to_workbook
-        if isinstance(df, dict) and len(df) > 1:
+        # if isinstance(df, dict) and len(df) > 1:
+        if isinstance(df, dict):
             return RAGExcelParser._dataframes_to_workbook(df)
 
         df = RAGExcelParser._clean_dataframe(df)

From 37f72f919f51f692fe6fc7be2a611f05b000b902 Mon Sep 17 00:00:00 2001
From: lixiangcheng1 <lixiangcheng1@wanda.cn>
Date: Tue, 30 Dec 2025 19:31:54 +0800
Subject: [PATCH 5/8] [fix]parsed excel document error:float division by zero

---
 api/app/core/rag/deepdoc/parser/excel_parser.py | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)

diff --git a/api/app/core/rag/deepdoc/parser/excel_parser.py b/api/app/core/rag/deepdoc/parser/excel_parser.py
index 07e39f4a..f7601ee3 100644
--- a/api/app/core/rag/deepdoc/parser/excel_parser.py
+++ b/api/app/core/rag/deepdoc/parser/excel_parser.py
@@ -42,35 +42,31 @@ class RAGExcelParser:
                 file_like_object.seek(0)
                 try:
                     dfs = pd.read_excel(file_like_object, sheet_name=None)
+                    if isinstance(dfs, dict):
+                        dfs = next(iter(dfs.values()))
                     return RAGExcelParser._dataframe_to_workbook(dfs)
                 except Exception as ex:
                     logging.info(f"pandas with default engine load error: {ex}, try calamine instead")
                     file_like_object.seek(0)
                     df = pd.read_excel(file_like_object, engine="calamine")
+                    print(df)
                     return RAGExcelParser._dataframe_to_workbook(df)
             except Exception as e_pandas:
                 raise Exception(f"pandas.read_excel error: {e_pandas}, original openpyxl error: {e}")
 
     @staticmethod
-    def _clean_dataframe(df):
+    def _clean_dataframe(df: pd.DataFrame):
         def clean_string(s):
             if isinstance(s, str):
                 return ILLEGAL_CHARACTERS_RE.sub(" ", s)
             return s
 
-        # 处理单 DataFrame 或字典（多 Sheet）
-        if isinstance(df, dict):
-            return {sheet: RAGExcelParser._clean_dataframe(sheet_df) for sheet, sheet_df in df.items()}
-        elif isinstance(df, pd.DataFrame):
-            return df.apply(lambda col: col.map(clean_string))
-        else:
-            raise ValueError(f"Unsupported type for cleaning: {type(df)}")
+        return df.apply(lambda col: col.map(clean_string))
 
     @staticmethod
     def _dataframe_to_workbook(df):
         # if contains multiple sheets use _dataframes_to_workbook
-        # if isinstance(df, dict) and len(df) > 1:
-        if isinstance(df, dict):
+        if isinstance(df, dict) and len(df) > 1:
             return RAGExcelParser._dataframes_to_workbook(df)
 
         df = RAGExcelParser._clean_dataframe(df)

From 07bcb54ed3320742cab0e469096899a0c2c1f031 Mon Sep 17 00:00:00 2001
From: lixiangcheng1 <lixiangcheng1@wanda.cn>
Date: Wed, 31 Dec 2025 08:54:36 +0800
Subject: [PATCH 6/8] [fix]entity_resolution.py:199: SyntaxWarning: invalid
 escape sequence '\d'

---
 api/app/core/rag/graphrag/entity_resolution.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/api/app/core/rag/graphrag/entity_resolution.py b/api/app/core/rag/graphrag/entity_resolution.py
index d648e5a1..20af8818 100644
--- a/api/app/core/rag/graphrag/entity_resolution.py
+++ b/api/app/core/rag/graphrag/entity_resolution.py
@@ -196,7 +196,7 @@ class EntityResolution(Extractor):
         ans_list = []
         records = [r.strip() for r in results.split(record_delimiter)]
         for record in records:
-            pattern_int = f"{re.escape(entity_index_delimiter)}(\d+){re.escape(entity_index_delimiter)}"
+            pattern_int = fr"{re.escape(entity_index_delimiter)}(\d+){re.escape(entity_index_delimiter)}"
             match_int = re.search(pattern_int, record)
             res_int = int(str(match_int.group(1) if match_int else '0'))
             if res_int > records_length:

From c78dc1fd4719cbb543b57b7cf69a6c397a083416 Mon Sep 17 00:00:00 2001
From: lixiangcheng1 <lixiangcheng1@wanda.cn>
Date: Wed, 31 Dec 2025 09:51:00 +0800
Subject: [PATCH 7/8] [fix]parsed excel document error:float division by zero

---
 api/app/core/rag/app/naive.py                 |  1 +
 .../core/rag/deepdoc/parser/excel_parser.py   | 77 +++++++++++++++----
 2 files changed, 61 insertions(+), 17 deletions(-)

diff --git a/api/app/core/rag/app/naive.py b/api/app/core/rag/app/naive.py
index 6d6b933a..23f0c4ba 100644
--- a/api/app/core/rag/app/naive.py
+++ b/api/app/core/rag/app/naive.py
@@ -672,6 +672,7 @@ def chunk(filename, binary=None, from_page=0, to_page=100000,
         excel_parser = ExcelParser()
         if parser_config.get("html4excel"):
             sections = [(_, "") for _ in excel_parser.html(binary, 12) if _]
+            parser_config["chunk_token_num"] = 0
         else:
             sections = [(_, "") for _ in excel_parser(binary) if _]
         parser_config["chunk_token_num"] = 12800
diff --git a/api/app/core/rag/deepdoc/parser/excel_parser.py b/api/app/core/rag/deepdoc/parser/excel_parser.py
index f7601ee3..a161f4ca 100644
--- a/api/app/core/rag/deepdoc/parser/excel_parser.py
+++ b/api/app/core/rag/deepdoc/parser/excel_parser.py
@@ -5,6 +5,7 @@ from io import BytesIO
 
 import pandas as pd
 from openpyxl import Workbook, load_workbook
+from PIL import Image
 
 from app.core.rag.nlp import find_codec
 
@@ -28,7 +29,7 @@ class RAGExcelParser:
 
             try:
                 file_like_object.seek(0)
-                df = pd.read_csv(file_like_object)
+                df = pd.read_csv(file_like_object, on_bad_lines='skip')
                 return RAGExcelParser._dataframe_to_workbook(df)
 
             except Exception as e_csv:
@@ -42,14 +43,12 @@ class RAGExcelParser:
                 file_like_object.seek(0)
                 try:
                     dfs = pd.read_excel(file_like_object, sheet_name=None)
-                    if isinstance(dfs, dict):
-                        dfs = next(iter(dfs.values()))
                     return RAGExcelParser._dataframe_to_workbook(dfs)
                 except Exception as ex:
                     logging.info(f"pandas with default engine load error: {ex}, try calamine instead")
                     file_like_object.seek(0)
                     df = pd.read_excel(file_like_object, engine="calamine")
-                    print(df)
+                    print("lxc1")
                     return RAGExcelParser._dataframe_to_workbook(df)
             except Exception as e_pandas:
                 raise Exception(f"pandas.read_excel error: {e_pandas}, original openpyxl error: {e}")
@@ -68,7 +67,6 @@ class RAGExcelParser:
         # if contains multiple sheets use _dataframes_to_workbook
         if isinstance(df, dict) and len(df) > 1:
             return RAGExcelParser._dataframes_to_workbook(df)
-
         df = RAGExcelParser._clean_dataframe(df)
         wb = Workbook()
         ws = wb.active
@@ -80,15 +78,14 @@ class RAGExcelParser:
         for row_num, row in enumerate(df.values, 2):
             for col_num, value in enumerate(row, 1):
                 ws.cell(row=row_num, column=col_num, value=value)
-
         return wb
-    
+
     @staticmethod
     def _dataframes_to_workbook(dfs: dict):
         wb = Workbook()
         default_sheet = wb.active
         wb.remove(default_sheet)
-        
+
         for sheet_name, df in dfs.items():
             df = RAGExcelParser._clean_dataframe(df)
             ws = wb.create_sheet(title=sheet_name)
@@ -99,6 +96,52 @@ class RAGExcelParser:
                     ws.cell(row=row_num, column=col_num, value=value)
         return wb
 
+    @staticmethod
+    def _extract_images_from_worksheet(ws, sheetname=None):
+        """
+        Extract images from a worksheet and enrich them with vision-based descriptions.
+
+        Returns: List[dict]
+        """
+        images = getattr(ws, "_images", [])
+        if not images:
+            return []
+
+        raw_items = []
+
+        for img in images:
+            try:
+                img_bytes = img._data()
+                pil_img = Image.open(BytesIO(img_bytes)).convert("RGB")
+
+                anchor = img.anchor
+                if hasattr(anchor, "_from") and hasattr(anchor, "_to"):
+                    r1, c1 = anchor._from.row + 1, anchor._from.col + 1
+                    r2, c2 = anchor._to.row + 1, anchor._to.col + 1
+                    if r1 == r2 and c1 == c2:
+                        span = "single_cell"
+                    else:
+                        span = "multi_cell"
+                else:
+                    r1, c1 = anchor._from.row + 1, anchor._from.col + 1
+                    r2, c2 = r1, c1
+                    span = "single_cell"
+
+                item = {
+                    "sheet": sheetname or ws.title,
+                    "image": pil_img,
+                    "image_description": "",
+                    "row_from": r1,
+                    "col_from": c1,
+                    "row_to": r2,
+                    "col_to": c2,
+                    "span_type": span,
+                }
+                raw_items.append(item)
+            except Exception:
+                continue
+        return raw_items
+
     def html(self, fnm, chunk_rows=256):
         from html import escape
 
@@ -131,7 +174,7 @@ class RAGExcelParser:
                 tb = ""
                 tb += f"<table><caption>{sheetname}</caption>"
                 tb += tb_rows_0
-                for r in list(rows[1 + chunk_i * chunk_rows : min(1 + (chunk_i + 1) * chunk_rows, len(rows))]):
+                for r in list(rows[1 + chunk_i * chunk_rows: min(1 + (chunk_i + 1) * chunk_rows, len(rows))]):
                     tb += "<tr>"
                     for i, c in enumerate(r):
                         if c.value is None:
@@ -154,7 +197,7 @@ class RAGExcelParser:
         except Exception as e:
             logging.warning(f"Parse spreadsheet error: {e}, trying to interpret as CSV file")
             file_like_object.seek(0)
-            df = pd.read_csv(file_like_object)
+            df = pd.read_csv(file_like_object, on_bad_lines='skip')
         df = df.replace(r"^\s*$", "", regex=True)
         return df.to_markdown(index=False)
 
@@ -192,14 +235,14 @@ class RAGExcelParser:
         if fnm.split(".")[-1].lower().find("xls") >= 0:
             wb = RAGExcelParser._load_excel_to_workbook(BytesIO(binary))
             total = 0
-            
+
             for sheetname in wb.sheetnames:
-               try:
-                   ws = wb[sheetname]
-                   total += len(list(ws.rows))
-               except Exception as e:
-                   logging.warning(f"Skip sheet '{sheetname}' due to rows access error: {e}")
-                   continue
+                try:
+                    ws = wb[sheetname]
+                    total += len(list(ws.rows))
+                except Exception as e:
+                    logging.warning(f"Skip sheet '{sheetname}' due to rows access error: {e}")
+                    continue
             return total
 
         if fnm.split(".")[-1].lower() in ["csv", "txt"]:

From 6b0ee1b74a079f32d5482b52ae8a0836b19ef917 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E6=9D=8E=E6=96=B0=E6=9C=88?=
 <accounts_68c0f5d519f260d93ee2997e@mail.teambition.com>
Date: Wed, 31 Dec 2025 02:20:45 +0000
Subject: [PATCH 8/8] Merge #88 into develop from fix/develop_kj_knowledge
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

提示词优化

* fix/develop_kj_knowledge: (5 commits squashed)

  - Agent应用中添加知识库的配置字段（提示词修改、反思给默认值）

  - 提示词优化

  - 提示词优化

  - 提示词优化

  - 提示词优化

Signed-off-by: aliyun8644380055 <accounts_68c0f5d519f260d93ee2997e@mail.teambition.com>
Reviewed-by: aliyun6762716068 <accounts_68cb7c6b61f5dcc4200d6251@mail.teambition.com>
Merged-by: aliyun6762716068 <accounts_68cb7c6b61f5dcc4200d6251@mail.teambition.com>

CR-link: https://codeup.aliyun.com/redbearai/python/redbear-mem-open/change/88
---
 .../reflection_engine/example/example.json    | 48 ++++---------------
 .../reflection_engine/self_reflexion.py       | 20 +++++++-
 .../utils/prompt/prompts/reflexion.jinja2     |  6 ++-
 3 files changed, 32 insertions(+), 42 deletions(-)

diff --git a/api/app/core/memory/storage_services/reflection_engine/example/example.json b/api/app/core/memory/storage_services/reflection_engine/example/example.json
index 18a2b185..09429829 100644
--- a/api/app/core/memory/storage_services/reflection_engine/example/example.json
+++ b/api/app/core/memory/storage_services/reflection_engine/example/example.json
@@ -2,52 +2,39 @@
   "memory_verify": {
     "source_data": [
       {
-        "statement_name": "用户是2023年春天去北京工作的。",
-        "statement_id": "62beac695b1346f4871740a45db88782"
+        "statement_name": "我是 2023 年春天去北京工作的，后来基本一直都在北京上班，也没怎么换过城市。不过后来公司调整，2024 年上半年我被调到上海待了差不多半年，那段时间每天都是在上海办公室打卡。当时入职资料用的还是我之前的身份信息，身份证号是 11010119950308123X，银行卡是 6222023847595898，这些一直没变。对了，其实我 从 2023 年开始就一直在北京生活，从来没有长期离开过北京，上海那段更多算是远程配合。"
       },
       {
-        "statement_name": "用户后来基本一直都在北京上班。",
-        "statement_id": "4cba5ac08b674d7fb1e2ae634d2b8f0b"
+        "statement_name": "用户后来基本一直都在北京上班。"
       },
       {
-        "statement_name": "用户从2023年开始就一直在北京生活。",
-        "statement_id": "e612a44da4db483993c350df7c97a1a1"
+        "statement_name": "用户从2023年开始就一直在北京生活。"
       },
       {
-        "statement_name": "用户从来没有长期离开过北京。",
-        "statement_id": "b3c787a2e33c49f7981accabbbb4538a"
+        "statement_name": "用户从来没有长期离开过北京。"
       },
       {
-        "statement_name": "由于公司调整，用户在2024年上半年被调到上海待了差不多半年。",
-        "statement_id": "64cde4230cb24a4da726e7db9e7aa616"
+        "statement_name": "由于公司调整，用户在2024年上半年被调到上海待了差不多半年。"
       },
       {
-        "statement_name": "用户在被调到上海期间每天都是在上海办公室打卡。",
-        "statement_id": "8b1b12e23b844b8088dfeb67da6ad669"
+        "statement_name": "用户在被调到上海期间每天都是在上海办公室打卡。"
       },
       {
-        "statement_name": "用户在入职时使用的身份信息是之前的，身份证号为11010119950308123X。",
-        "statement_id": "030afd362e9b4110b139e68e5d3e7143"
+        "statement_name": "用户在入职时使用的身份信息是之前的，身份证号为11010119950308123X。"
       },
       {
-        "statement_name": "用户的银行卡号是6222023847595898。",
-        "statement_id": "6c7567cd1f3c478bb42d1b65383e6f2f"
+        "statement_name": "用户的银行卡号是6222023847595898。"
       },
       {
-        "statement_name": "用户的身份信息和银行卡信息一直没变。",
-        "statement_id": "b3ca618e1e204b83bebd70e75cf2073f"
+        "statement_name": "用户的身份信息和银行卡信息一直没变。"
       },
       {
-        "statement_name": "用户认为在上海的那段时间更多算是远程配合。",
-        "statement_id": "150af89d2c154e6eb41ff1a91e37f962"
+        "statement_name": "用户认为在上海的那段时间更多算是远程配合。"
       }
     ],
     "databasets": [
       {
         "entity1_name": "Person",
-        "description": "表示人类个体的通用类型",
-        "statement_id": "62beac695b1346f4871740a45db88782",
-        "entity2_name": "用户",
         "entity2": {
           "description": "叙述者，讲述个人工作与生活经历的个体",
           "name": "用户"
@@ -55,9 +42,6 @@
       },
       {
         "entity1_name": "用户",
-        "description": "叙述者，讲述个人工作与生活经历的个体",
-        "statement_id": "62beac695b1346f4871740a45db88782",
-        "entity2_name": "身份信息",
         "entity2": {
           "description": "用于个人身份识别的数据",
           "name": "身份信息"
@@ -65,9 +49,6 @@
       },
       {
         "entity1_name": "用户",
-        "description": "叙述者，讲述个人工作与生活经历的个体",
-        "statement_id": "62beac695b1346f4871740a45db88782",
-        "entity2_name": "6222023847595898",
         "entity2": {
           "description": "用户的银行卡号码",
           "name": "6222023847595898"
@@ -76,33 +57,24 @@
       {
         "entity1_name": "用户",
         "description": "叙述者，讲述个人工作与生活经历的个体",
-        "statement_id": "62beac695b1346f4871740a45db88782",
-        "entity2_name": "上海办公室",
         "entity2": {
           "entity_idx": 1,
           "aliases": ["上海办"],
-          "description": "位于上海的工作办公场所",
           "name": "上海办公室"
         }
       },
       {
         "entity1_name": "用户",
         "description": "叙述者，讲述个人工作与生活经历的个体",
-        "statement_id": "62beac695b1346f4871740a45db88782",
-        "entity2_name": "北京",
         "entity2": {
           "aliases": ["京", "京城", "北平"],
-          "description": "中国的首都城市，用户主要工作和生活所在地",
           "name": "北京"
         }
       },
       {
         "entity1_name": "11010119950308123X",
         "description": "具体的身份证号码值",
-        "statement_id": "030afd362e9b4110b139e68e5d3e7143",
-        "entity2_name": "身份证号",
         "entity2": {
-          "description": "中华人民共和国公民的身份号码",
           "name": "身份证号"
         }
       }
diff --git a/api/app/core/memory/storage_services/reflection_engine/self_reflexion.py b/api/app/core/memory/storage_services/reflection_engine/self_reflexion.py
index 97f51fb9..e9fb8855 100644
--- a/api/app/core/memory/storage_services/reflection_engine/self_reflexion.py
+++ b/api/app/core/memory/storage_services/reflection_engine/self_reflexion.py
@@ -387,7 +387,7 @@ class ReflectionEngine:
         result_data['memory_verifies'] = memory_verifies
         result_data['quality_assessments'] = quality_assessments
         conflicts_found=''
-
+        REMOVE_KEYS = {"created_at", "expired_at","relationship","predicate","statement_id","id","statement_id","relationship_statement_id"}
         # Clearn conflict_data，And memory_verify和quality_assessment
         cleaned_conflict_data = []
         for item in conflict_data:
@@ -396,7 +396,23 @@ class ReflectionEngine:
                 'conflict': item['conflict']
             }
             cleaned_conflict_data.append(cleaned_item)
-
+        cleaned_conflict_data_=[]
+        for item in conflict_data:
+            cleaned_data = []
+            for row in item.get("data", []):
+                # 删除 created_at / expired_at
+                cleaned_row = {
+                    k: v
+                    for k, v in row.items()
+                    if k not in REMOVE_KEYS
+                }
+                cleaned_data.append(cleaned_row)
+            cleaned_item = {
+                "data": cleaned_data,
+                "conflict": item.get("conflict"),
+            }
+            cleaned_conflict_data_.append(cleaned_item)
+        print(cleaned_conflict_data_)
         # 3. 解决冲突
         solved_data = await self._resolve_conflicts(cleaned_conflict_data, source_data)
         if not solved_data:
diff --git a/api/app/core/memory/utils/prompt/prompts/reflexion.jinja2 b/api/app/core/memory/utils/prompt/prompts/reflexion.jinja2
index 99476c82..91b067ee 100644
--- a/api/app/core/memory/utils/prompt/prompts/reflexion.jinja2
+++ b/api/app/core/memory/utils/prompt/prompts/reflexion.jinja2
@@ -111,7 +111,8 @@
 - 隐私保护优先: 所有输出记录必须完成隐私脱敏
 - 脱敏变更记录: 隐私脱敏变更也必须在change字段中记录{% endif %}
 - 不可修改数据: 数据被判定为正确时不可修改，无数据可输出时为空
-- 输出的结果reflexion字段中的reason字段和solution不允许含有（expired_at设为2024-01-01T00:00:00Z、memory_verify=true)等原数据字段以及涉及需要修改的字段以及内容
+- 输出的结果reflexion字段中的reason字段和solution不允许含有（expired_at设为2024-01-01T00:00:00Z、memory_verify=true、memory_verify=false)等原数据字段以及涉及需要修改的字段以及内容，
+    ，如果是FACT，只记录事实冲突相关的数据；如果是TIME，只记录时间冲突相关的数据；如果是HYBRID，则记录所有冲突相关的数据
 
 **变更记录格式**:
 ```json
@@ -158,7 +159,8 @@
         "conflict": true
       },
       "reflexion": {
-        "reason": "该冲突类型的原因分析",
+        "reason": "该冲突类型的原因分析，如果是FACT就是存在事实冲突，分析该冲突原因，如果是TIME就是存在时间冲突，分析该冲突原因，如果是HYBRID，可以输出存在时间与事实的混合冲突再添加上原因分析，
+        不可以随意分配冲突类型以及原因",
         "solution": "该冲突类型的解决方案"
       },
       "resolved": {