diff --git a/agent/component/base.py b/agent/component/base.py
index 81d3fac56..321907dbe 100644
--- a/agent/component/base.py
+++ b/agent/component/base.py
@@ -393,7 +393,7 @@ class ComponentParamBase(ABC):
 class ComponentBase(ABC):
     component_name: str
     thread_limiter = asyncio.Semaphore(int(os.environ.get("MAX_CONCURRENT_CHATS", 10)))
-    variable_ref_patt = r"\{* *\{([a-zA-Z:0-9]+@[A-Za-z0-9_.]+|sys\.[A-Za-z0-9_.]+|env\.[A-Za-z0-9_.]+)\} *\}*"
+    variable_ref_patt = r"\{* *\{([a-zA-Z_:0-9]+@[A-Za-z0-9_.]+|sys\.[A-Za-z0-9_.]+|env\.[A-Za-z0-9_.]+)\} *\}*"
 
     def __str__(self):
         """
diff --git a/agent/component/message.py b/agent/component/message.py
index b4e2985e0..e2428a084 100644
--- a/agent/component/message.py
+++ b/agent/component/message.py
@@ -14,6 +14,8 @@
 #  limitations under the License.
 #
 import asyncio
+import nest_asyncio
+nest_asyncio.apply()
 import inspect
 import json
 import os
@@ -207,7 +209,7 @@ class Message(ComponentBase):
         import pypandoc
         doc_id = get_uuid()
 
-        if self._param.output_format.lower() not in {"markdown", "html", "pdf", "docx"}:
+        if self._param.output_format.lower() not in {"markdown", "html", "pdf", "docx", "xlsx"}:
             self._param.output_format = "markdown"
 
         try:
@@ -227,6 +229,46 @@ class Message(ComponentBase):
 
                 binary_content = converted.encode("utf-8")
 
+            elif self._param.output_format == "xlsx":
+                import pandas as pd
+                from io import BytesIO
+
+                if isinstance(content, str):
+                    try:
+                        # Convert markdown to HTML tables to help pandas parse it
+                        html_content = pypandoc.convert_text(content, to="html", format="markdown")
+                        dfs = pd.read_html(html_content)
+                    except Exception as e:
+                        dfs = []
+                    
+                    if not dfs:
+                        df = pd.DataFrame({"Content": [content]})
+                        dfs = [df]
+                else:
+                    # Should not accept file path for Excel generation from agent response usually, 
+                    # but if it does, read it as text
+                    with open(content, "r") as f:
+                        txt_content = f.read()
+                    try:
+                         html_content = pypandoc.convert_text(txt_content, to="html", format="markdown")
+                         dfs = pd.read_html(html_content)
+                    except Exception:
+                        dfs = []
+                    
+                    if not dfs:
+                        df = pd.DataFrame({"Content": [txt_content]})
+                        dfs = [df]
+
+                # Write to Excel
+                excel_io = BytesIO()
+                with pd.ExcelWriter(excel_io, engine='openpyxl') as writer:
+                    for i, df in enumerate(dfs):
+                        sheet_name = f"Sheet{i+1}"
+                        df.to_excel(writer, sheet_name=sheet_name, index=False)
+                
+                excel_io.seek(0)
+                binary_content = excel_io.read()
+
             else:  # pdf, docx
                 with tempfile.NamedTemporaryFile(suffix=f".{self._param.output_format}", delete=False) as tmp:
                     tmp_name = tmp.name
diff --git a/api/apps/canvas_app.py b/api/apps/canvas_app.py
index ed8c8c7a0..64b0d0f55 100644
--- a/api/apps/canvas_app.py
+++ b/api/apps/canvas_app.py
@@ -14,6 +14,7 @@
 #  limitations under the License.
 #
 import asyncio
+import inspect
 import json
 import logging
 from functools import partial
@@ -299,8 +300,13 @@ async def debug():
         for k in outputs.keys():
             if isinstance(outputs[k], partial):
                 txt = ""
-                for c in outputs[k]():
-                    txt += c
+                iter_obj = outputs[k]()
+                if inspect.isasyncgen(iter_obj):
+                    async for c in iter_obj:
+                        txt += c
+                else:
+                    for c in iter_obj:
+                        txt += c
                 outputs[k] = txt
         return get_json_result(data=outputs)
     except Exception as e:
diff --git a/web/src/pages/agent/constant/index.tsx b/web/src/pages/agent/constant/index.tsx
index 5c25b7fe0..8775c6288 100644
--- a/web/src/pages/agent/constant/index.tsx
+++ b/web/src/pages/agent/constant/index.tsx
@@ -832,6 +832,7 @@ export enum ExportFileType {
   HTML = 'html',
   Markdown = 'md',
   DOCX = 'docx',
+  Excel = 'xlsx',
 }
 
 export enum TypesWithArray {