diff --git a/llama_stack/providers/inline/agents/meta_reference/agent_instance.py b/llama_stack/providers/inline/agents/meta_reference/agent_instance.py
index 8c10c6baf..4a1421245 100644
--- a/llama_stack/providers/inline/agents/meta_reference/agent_instance.py
+++ b/llama_stack/providers/inline/agents/meta_reference/agent_instance.py
@@ -909,7 +909,19 @@ class ChatAgent(ShieldRunnerMixin):
                     if tool_def_map.get(built_in_type, None):
                         raise ValueError(f"Tool {built_in_type} already exists")
 
-                    tool_def_map[built_in_type] = ToolDefinition(tool_name=built_in_type)
+                    tool_def_map[built_in_type] = ToolDefinition(
+                        tool_name=built_in_type,
+                        description=tool_def.description,
+                        parameters={
+                            param.name: ToolParamDefinition(
+                                param_type=param.parameter_type,
+                                description=param.description,
+                                required=param.required,
+                                default=param.default,
+                            )
+                            for param in tool_def.parameters
+                        },
+                    )
                     tool_to_group[built_in_type] = tool_def.toolgroup_id
                     continue
 
diff --git a/llama_stack/providers/remote/inference/vllm/vllm.py b/llama_stack/providers/remote/inference/vllm/vllm.py
index b9422d85d..967a3e44d 100644
--- a/llama_stack/providers/remote/inference/vllm/vllm.py
+++ b/llama_stack/providers/remote/inference/vllm/vllm.py
@@ -169,7 +169,7 @@ async def _process_vllm_chat_completion_stream_response(
                 args = {} if not args_str else json.loads(args_str)
             except Exception as e:
                 log.warning(f"Failed to parse tool call buffer arguments: {args_str} \nError: {e}")
-            if args is not None:
+            if args:
                 yield ChatCompletionResponseStreamChunk(
                     event=ChatCompletionResponseEvent(
                         event_type=event_type,
@@ -183,7 +183,7 @@ async def _process_vllm_chat_completion_stream_response(
                         ),
                     )
                 )
-            else:
+            elif args_str:
                 yield ChatCompletionResponseStreamChunk(
                     event=ChatCompletionResponseEvent(
                         event_type=ChatCompletionResponseEventType.progress,