diff --git a/llama_stack/providers/inline/agents/meta_reference/responses/streaming.py b/llama_stack/providers/inline/agents/meta_reference/responses/streaming.py
index efe06cecf..0a7ef6065 100644
--- a/llama_stack/providers/inline/agents/meta_reference/responses/streaming.py
+++ b/llama_stack/providers/inline/agents/meta_reference/responses/streaming.py
@@ -264,6 +264,7 @@ class StreamingResponseOrchestrator:
                         completion_result_data = stream_event_or_result
                     else:
                         yield stream_event_or_result
+
                 # If violation detected, skip the rest of processing since we already sent refusal
                 if self.violation_detected:
                     return
diff --git a/llama_stack/providers/inline/agents/meta_reference/responses/utils.py b/llama_stack/providers/inline/agents/meta_reference/responses/utils.py
index e55e7b1fb..7d6ac1faa 100644
--- a/llama_stack/providers/inline/agents/meta_reference/responses/utils.py
+++ b/llama_stack/providers/inline/agents/meta_reference/responses/utils.py
@@ -174,7 +174,7 @@ async def convert_response_input_to_chat_messages(
                 pass
             else:
                 content = await convert_response_content_to_chat_content(input_item.content)
-                message_type = get_message_type_by_role(input_item.role)
+                message_type = await get_message_type_by_role(input_item.role)
                 if message_type is None:
                     raise ValueError(
                         f"Llama Stack OpenAI Responses does not yet support message role '{input_item.role}' in this context"
diff --git a/tests/unit/providers/agents/meta_reference/test_responses_safety_utils.py b/tests/unit/providers/agents/meta_reference/test_responses_safety_utils.py
index 1df1d0d82..b5abf67d0 100644
--- a/tests/unit/providers/agents/meta_reference/test_responses_safety_utils.py
+++ b/tests/unit/providers/agents/meta_reference/test_responses_safety_utils.py
@@ -83,7 +83,7 @@ def test_extract_shield_ids_unknown_format(responses_impl):
     # Create an object that's neither string nor ResponseShieldSpec
     unknown_object = {"invalid": "format"}  # Plain dict, not ResponseShieldSpec
     shields = ["valid-shield", unknown_object, "another-shield"]
-    with pytest.raises(ValueError, match="Unsupported shield type"):
+    with pytest.raises(ValueError, match="Unknown shield format.*expected str or ResponseShieldSpec"):
         extract_shield_ids(shields)