fix: make sure system message is present in reflection call (#5926)

Resolves #5919

python/packages/autogen-agentchat/src/autogen_agentchat/agents/_assistant_agent.py

@@ -784,6 +784,7 @@ class AssistantAgent(BaseChatAgent, Component[AssistantAgentConfig]):
             inner_messages=inner_messages,
             cancellation_token=cancellation_token,
             agent_name=agent_name,
+            system_messages=system_messages,
             model_context=model_context,
             tools=tools,
             handoff_tools=handoff_tools,
@@ -878,6 +879,7 @@ class AssistantAgent(BaseChatAgent, Component[AssistantAgentConfig]):
         inner_messages: List[AgentEvent | ChatMessage],
         cancellation_token: CancellationToken,
         agent_name: str,
+        system_messages: List[SystemMessage],
         model_context: ChatCompletionContext,
         tools: List[BaseTool[Any, Any]],
         handoff_tools: List[BaseTool[Any, Any]],
@@ -959,6 +961,7 @@ class AssistantAgent(BaseChatAgent, Component[AssistantAgentConfig]):
         # STEP 4D: Reflect or summarize tool results
         if reflect_on_tool_use:
             async for reflection_response in AssistantAgent._reflect_on_tool_use_flow(
+                system_messages=system_messages,
                 model_client=model_client,
                 model_client_stream=model_client_stream,
                 model_context=model_context,
@@ -1039,6 +1042,7 @@ class AssistantAgent(BaseChatAgent, Component[AssistantAgentConfig]):
     @classmethod
     async def _reflect_on_tool_use_flow(
         cls,
+        system_messages: List[SystemMessage],
         model_client: ChatCompletionClient,
         model_client_stream: bool,
         model_context: ChatCompletionContext,
@@ -1049,7 +1053,7 @@ class AssistantAgent(BaseChatAgent, Component[AssistantAgentConfig]):
         If reflect_on_tool_use=True, we do another inference based on tool results
         and yield the final text response (or streaming chunks).
         """
-        all_messages = await model_context.get_messages()
+        all_messages = system_messages + await model_context.get_messages()
         llm_messages = cls._get_compatible_context(model_client=model_client, messages=all_messages)
 
         reflection_result: Optional[CreateResult] = None

python/packages/autogen-agentchat/tests/test_assistant_agent.py

@@ -25,6 +25,7 @@ from autogen_core.models import (
     AssistantMessage,
     CreateResult,
     FunctionExecutionResult,
+    FunctionExecutionResultMessage,
     LLMMessage,
     RequestUsage,
     SystemMessage,
@@ -80,6 +81,15 @@ async def test_run_with_tools(monkeypatch: pytest.MonkeyPatch) -> None:
     )
     result = await agent.run(task="task")
 
+    # Make sure the create call was made with the correct parameters.
+    assert len(model_client.create_calls) == 1
+    llm_messages = model_client.create_calls[0]["messages"]
+    assert len(llm_messages) == 2
+    assert isinstance(llm_messages[0], SystemMessage)
+    assert llm_messages[0].content == agent._system_messages[0].content  # type: ignore
+    assert isinstance(llm_messages[1], UserMessage)
+    assert llm_messages[1].content == "task"
+
     assert len(result.messages) == 5
     assert isinstance(result.messages[0], TextMessage)
     assert result.messages[0].models_usage is None
@@ -150,6 +160,23 @@ async def test_run_with_tools_and_reflection() -> None:
     )
     result = await agent.run(task="task")
 
+    # Make sure the create call was made with the correct parameters.
+    assert len(model_client.create_calls) == 2
+    llm_messages = model_client.create_calls[0]["messages"]
+    assert len(llm_messages) == 2
+    assert isinstance(llm_messages[0], SystemMessage)
+    assert llm_messages[0].content == agent._system_messages[0].content  # type: ignore
+    assert isinstance(llm_messages[1], UserMessage)
+    assert llm_messages[1].content == "task"
+    llm_messages = model_client.create_calls[1]["messages"]
+    assert len(llm_messages) == 4
+    assert isinstance(llm_messages[0], SystemMessage)
+    assert llm_messages[0].content == agent._system_messages[0].content  # type: ignore
+    assert isinstance(llm_messages[1], UserMessage)
+    assert llm_messages[1].content == "task"
+    assert isinstance(llm_messages[2], AssistantMessage)
+    assert isinstance(llm_messages[3], FunctionExecutionResultMessage)
+
     assert len(result.messages) == 4
     assert isinstance(result.messages[0], TextMessage)
     assert result.messages[0].models_usage is None