diff --git a/llama_stack/providers/utils/inference/openai_compat.py b/llama_stack/providers/utils/inference/openai_compat.py
index ac9d18312..8ee838d84 100644
--- a/llama_stack/providers/utils/inference/openai_compat.py
+++ b/llama_stack/providers/utils/inference/openai_compat.py
@@ -175,6 +175,7 @@ def process_chat_completion_response(
     choice = response.choices[0]
 
     # TODO: This does not work well with tool calls for vLLM remote provider
+    #   Ref: https://github.com/meta-llama/llama-stack/issues/1058
     raw_message = formatter.decode_assistant_message_from_content(
         text_from_choice(choice), get_stop_reason(choice.finish_reason)
     )