diff --git a/agentflow_cli/src/app/routers/graph/services/graph_service.py b/agentflow_cli/src/app/routers/graph/services/graph_service.py
index acf922c..441b04d 100644
--- a/agentflow_cli/src/app/routers/graph/services/graph_service.py
+++ b/agentflow_cli/src/app/routers/graph/services/graph_service.py
@@ -295,7 +295,8 @@ async def stream_graph(
                 mt = chunk.metadata or {}
                 mt.update(meta)
                 chunk.metadata = mt
-                yield chunk.model_dump_json()
+                # Add newline delimiter and encode to bytes for proper chunk-by-chunk streaming
+                yield (chunk.model_dump_json() + "\n")
                 if (
                     self.config.thread_name_generator_path
                     and meta["is_new_thread"]
@@ -303,22 +304,32 @@ async def stream_graph(
                     and chunk.message
                     and not chunk.message.delta
                 ):
-                    messages_str.append(chunk.message.text())
+                    # Safely extract text - handle both string and Message object
+                    msg_text = (
+                        chunk.message.text()
+                        if hasattr(chunk.message, "text") and callable(chunk.message.text)
+                        else str(chunk.message)
+                    )
+                    messages_str.append(msg_text)
 
             logger.info("Graph streaming completed successfully")
 
             if meta["is_new_thread"] and self.config.thread_name_generator_path:
-                messages_str = [msg.text() for msg in messages_str]
+                # messages_str already contains text strings, no need to call .text() again
                 thread_name = await self._save_thread_name(
                     config, config["thread_id"], messages_str
                 )
                 meta["thread_name"] = thread_name
 
-                yield StreamChunk(
-                    event=StreamEvent.UPDATES,
-                    data={"status": "completed"},
-                    metadata=meta,
-                ).model_dump_json()
+                # Add newline delimiter and encode to bytes for proper chunk-by-chunk streaming
+                yield (
+                    StreamChunk(
+                        event=StreamEvent.UPDATES,
+                        data={"status": "completed"},
+                        metadata=meta,
+                    ).model_dump_json()
+                    + "\n"
+                )
 
         except Exception as e:
             logger.error(f"Graph streaming failed: {e}")
@@ -421,10 +432,10 @@ async def fix_graph(
                 "removed_count": 1,
                 "state": state.model_dump_json(),
             }
-        else:
-            logger.warning(
-                "Last message is not an assistant message with tool calls, skipping it from checks."
-            )
+
+        logger.warning(
+            "Last message is not an assistant message with tool calls, skipping it from checks."
+        )
 
         return {
             "success": True,
diff --git a/graph/react.py b/graph/react.py
index 7dd5c38..d84d7a9 100644
--- a/graph/react.py
+++ b/graph/react.py
@@ -70,44 +70,41 @@ class MyAgentState(AgentState):
 checkpointer = InMemoryCheckpointer[MyAgentState]()
 
 
-"""
-Note: The docstring below will be used as the tool description and it will be
-passed to the AI model for tool selection, so keep it relevant and concise.
-This function will be converted to a tool with the following schema:
-[
-        {
-            'type': 'function',
-            'function': {
-                'name': 'get_weather',
-                'description': 'Retrieve current weather information for a specified location.',
-                'parameters': {
-                    'type': 'object',
-                    'properties': {
-                        'location': {'type': 'string'}
-                    },
-                    'required': ['location']
-                }
-            }
-        }
-    ]
-
-Parameters like tool_call_id, state, and checkpointer are injected automatically
-by InjectQ when the tool is called by the agent.
-Available injected parameters:
-The following parameters are automatically injected by InjectQ when the tool is called,
-but need to keep them as same name and type for proper injection:
-- tool_call_id: Unique ID for the tool call
-- state: Current AgentState containing conversation context
-- config: Configuration dictionary passed during graph invocation
-
-Below fields need to be used with Inject[] to get the instances:
-- context_manager: ContextManager instance for managing context, like trimming
-- publisher: Publisher instance for publishing events and logs
-- checkpointer: InMemoryCheckpointer instance for state management
-- store: InMemoryStore instance for temporary data storage
-- callback: CallbackManager instance for handling callbacks
-
-"""
+# Note: The docstring below will be used as the tool description and it will be
+# passed to the AI model for tool selection, so keep it relevant and concise.
+# This function will be converted to a tool with the following schema:
+# [
+#         {
+#             'type': 'function',
+#             'function': {
+#                 'name': 'get_weather',
+#                 'description': 'Retrieve current weather information for a specified location.',
+#                 'parameters': {
+#                     'type': 'object',
+#                     'properties': {
+#                         'location': {'type': 'string'}
+#                     },
+#                     'required': ['location']
+#                 }
+#             }
+#         }
+#     ]
+#
+# Parameters like tool_call_id, state, and checkpointer are injected automatically
+# by InjectQ when the tool is called by the agent.
+# Available injected parameters:
+# The following parameters are automatically injected by InjectQ when the tool is called,
+# but need to keep them as same name and type for proper injection:
+# - tool_call_id: Unique ID for the tool call
+# - state: Current AgentState containing conversation context
+# - config: Configuration dictionary passed during graph invocation
+#
+# Below fields need to be used with Inject[] to get the instances:
+# - context_manager: ContextManager instance for managing context, like trimming
+# - publisher: Publisher instance for publishing events and logs
+# - checkpointer: InMemoryCheckpointer instance for state management
+# - store: InMemoryStore instance for temporary data storage
+# - callback: CallbackManager instance for handling callbacks
 
 
 def get_weather(