diff --git a/tools/sql_execute.py b/tools/sql_execute.py
index 76ba61d..2d4a2c1 100644
--- a/tools/sql_execute.py
+++ b/tools/sql_execute.py
@@ -2,13 +2,12 @@
 from typing import Any
 import re
 import json
-
+import pandas as pd
 import records
 from sqlalchemy import text
 from dify_plugin import Tool
 from dify_plugin.entities.tool import ToolInvokeMessage
 
-
 class SQLExecuteTool(Tool):
     def _invoke(self, tool_parameters: dict[str, Any]) -> Generator[ToolInvokeMessage]:
         db_uri = tool_parameters.get("db_uri") or self.runtime.credentials.get("db_uri")
@@ -25,38 +24,60 @@ def _invoke(self, tool_parameters: dict[str, Any]) -> Generator[ToolInvokeMessag
 
         try:
             if re.match(r'^\s*(SELECT|WITH)\s+', query, re.IGNORECASE):
+                # 查询字段类型
+                table_name = re.search(r'FROM\s+([^\s;]+)', query, re.IGNORECASE)
+                column_types = {}
+                if table_name:
+                    type_query = f"""
+                    SELECT COLUMN_NAME, DATA_TYPE, CHARACTER_MAXIMUM_LENGTH
+                    FROM INFORMATION_SCHEMA.COLUMNS
+                    WHERE TABLE_NAME = '{table_name.group(1)}'
+                    """
+                    types_rows = db.query(type_query)
+                    column_types = {row['COLUMN_NAME']: {
+                        'DATA_TYPE': row['DATA_TYPE'],
+                        'CHARACTER_MAXIMUM_LENGTH': row['CHARACTER_MAXIMUM_LENGTH']
+                    } for row in types_rows.as_dict()}
+
+                # 执行主查询
                 rows = db.query(query)
+                # 转换为 pandas DataFrame 并处理 NaN
+                df = pd.DataFrame(rows.as_dict())
+                df = df.fillna('')  # 替换 NaN 为空字符串
+
                 if format == "json":
-                    result = rows.as_dict()
-                    yield self.create_json_message({"result": result})
+                    result = df.to_dict(orient='records')
+                    yield self.create_json_message({"result": result, "column_types": column_types})
                 elif format == "md":
                     result = str(rows.dataset)
                     yield self.create_text_message(result)
                 elif format == "csv":
-                    result = rows.export("csv").encode()
+                    result = df.to_csv(index=False).encode()
                     yield self.create_blob_message(
                         result, meta={"mime_type": "text/csv", "filename": "result.csv"}
                     )
                 elif format == "yaml":
-                    result = rows.export("yaml").encode()
+                    result = df.to_dict(orient='records')
+                    result = json.dumps(result, ensure_ascii=False).encode()
                     yield self.create_blob_message(
-                        result,
-                        meta={"mime_type": "text/yaml", "filename": "result.yaml"},
+                        result, meta={"mime_type": "text/yaml", "filename": "result.yaml"}
                     )
                 elif format == "xlsx":
-                    result = rows.export("xlsx")
+                    output_file = '/tmp/output.xlsx'
+                    df.to_excel(output_file, index=False)
+                    with open(output_file, 'rb') as f:
+                        result = f.read()
                     yield self.create_blob_message(
                         result,
                         meta={
                             "mime_type": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
-                            "filename": "result.xlsx",
-                        },
+                            "filename": "result.xlsx"
+                        }
                     )
                 elif format == "html":
-                    result = rows.export("html").encode()
+                    result = df.to_html(index=False).encode()
                     yield self.create_blob_message(
-                        result,
-                        meta={"mime_type": "text/html", "filename": "result.html"},
+                        result, meta={"mime_type": "text/html", "filename": "result.html"}
                     )
                 else:
                     raise ValueError(f"Unsupported format: {format}")