diff --git a/llama_toolchain/agentic_system/client_sdk.py b/llama_toolchain/agentic_system/client_sdk.py
index f4c68d717..f8eb5550d 100644
--- a/llama_toolchain/agentic_system/client_sdk.py
+++ b/llama_toolchain/agentic_system/client_sdk.py
@@ -7,17 +7,6 @@ def main(host: str, port: int):
         base_url=f"http://{host}:{port}",
     )
 
-    # Need smt like this to work w/ server, however this is not what was generated by SDK
-    # response = client.inference.chat_completion(
-    #     request={
-    #         "messages": [
-    #             UserMessage(content="hello world, troll me in two-paragraphs about 42", role="user"),
-    #         ],
-    #         "model": "Meta-Llama3.1-8B-Instruct",
-    #         "stream": True,
-    #     },
-    # )
-
     agentic_system_create_response = client.agentic_system.create(
         agent_config={
             "instructions": "You are a helpful assistant",
@@ -32,7 +21,7 @@ def main(host: str, port: int):
     )
     print(agentic_system_create_session_response)
 
-    # TODO(xiyan): This does not work with current server, need to wrap it in a request (similar to AgentConfig?)
+    # TODO(xiyan): remove request wrapper
     response = client.agentic_system.turns.create(
         request={
             "agent_id": agentic_system_create_response.agent_id,
diff --git a/llama_toolchain/inference/client_sdk.py b/llama_toolchain/inference/client_sdk.py
index e0cf77f78..3abcd5d51 100644
--- a/llama_toolchain/inference/client_sdk.py
+++ b/llama_toolchain/inference/client_sdk.py
@@ -7,7 +7,6 @@ def main(host: str, port: int):
         base_url=f"http://{host}:{port}",
     )
 
-    # Need smt like this to work w/ server, however this is not what was generated by SDK (?)
     response = client.inference.chat_completion(
         request={
             "messages": [
@@ -19,7 +18,7 @@ def main(host: str, port: int):
     )
 
     print(response)
-    # This does not work with current server
+    # TODO (xiyan). This does not work with current server, need to fix
     # response = client.inference.chat_completion(
     #     messages=[
     #         UserMessage(content="hello world, troll me in two-paragraphs about 42", role="user"),