Merge pull request #17803 from BerriAI/litellm_preserve_system_instructions

fix: Preserve systemInstructions for vertex ai generate content request

litellm/google_genai/main.py

@@ -164,12 +164,15 @@ class GenerateContentHelper:
                 model=model,
             )
         )
+        # Extract systemInstruction from kwargs to pass to transform
+        system_instruction = kwargs.get("systemInstruction") or kwargs.get("system_instruction")
         request_body = (
             generate_content_provider_config.transform_generate_content_request(
                 model=model,
                 contents=contents,
                 tools=tools,
                 generate_content_config_dict=generate_content_config_dict,
+                system_instruction=system_instruction,
             )
         )
 
@@ -311,6 +314,9 @@ def generate_content(
             **kwargs,
         )
 
+        # Extract systemInstruction from kwargs to pass to handler
+        system_instruction = kwargs.get("systemInstruction") or kwargs.get("system_instruction")
+
         # Check if we should use the adapter (when provider config is None)
         if setup_result.generate_content_provider_config is None:
             # Use the adapter to convert to completion format
@@ -340,6 +346,7 @@ def generate_content(
             _is_async=_is_async,
             client=kwargs.get("client"),
             litellm_metadata=kwargs.get("litellm_metadata", {}),
+            system_instruction=system_instruction,
         )
 
         return response
@@ -395,6 +402,9 @@ async def agenerate_content_stream(
             **kwargs,
         )
 
+        # Extract systemInstruction from kwargs to pass to handler
+        system_instruction = kwargs.get("systemInstruction") or kwargs.get("system_instruction")
+
         # Check if we should use the adapter (when provider config is None)
         if setup_result.generate_content_provider_config is None:
             # Use the adapter to convert to completion format
@@ -428,6 +438,7 @@ async def agenerate_content_stream(
             client=kwargs.get("client"),
             stream=True,
             litellm_metadata=kwargs.get("litellm_metadata", {}),
+            system_instruction=system_instruction,
         )
 
     except Exception as e:

litellm/llms/base_llm/google_genai/transformation.py

@@ -149,6 +149,7 @@ class BaseGoogleGenAIGenerateContentConfig(ABC):
         contents: GenerateContentContentListUnionDict,
         tools: Optional[ToolConfigDict],
         generate_content_config_dict: Dict,
+        system_instruction: Optional[Any] = None,
     ) -> dict:
         """
         Transform the request parameters for the generate content API.
@@ -157,9 +158,8 @@ class BaseGoogleGenAIGenerateContentConfig(ABC):
             model: The model name
             contents: Input contents
             tools: Tools
-            generate_content_request_params: Request parameters
-            litellm_params: LiteLLM parameters
-            headers: Request headers
+            generate_content_config_dict: Generation config parameters
+            system_instruction: Optional system instruction
 
         Returns:
             Transformed request data

litellm/llms/custom_httpx/llm_http_handler.py

@@ -7311,6 +7311,7 @@ class BaseLLMHTTPHandler:
         client: Optional[Union[HTTPHandler, AsyncHTTPHandler]] = None,
         stream: bool = False,
         litellm_metadata: Optional[Dict[str, Any]] = None,
+        system_instruction: Optional[Any] = None,
     ) -> Any:
         """
         Handles Google GenAI generate content requests.
@@ -7336,6 +7337,7 @@ class BaseLLMHTTPHandler:
                 client=client if isinstance(client, AsyncHTTPHandler) else None,
                 stream=stream,
                 litellm_metadata=litellm_metadata,
+                system_instruction=system_instruction,
             )
 
         if client is None or not isinstance(client, HTTPHandler):
@@ -7365,6 +7367,7 @@ class BaseLLMHTTPHandler:
             contents=contents,
             tools=tools,
             generate_content_config_dict=generate_content_config_dict,
+            system_instruction=system_instruction,
         )
 
         if extra_body:
@@ -7435,6 +7438,7 @@ class BaseLLMHTTPHandler:
         client: Optional[AsyncHTTPHandler] = None,
         stream: bool = False,
         litellm_metadata: Optional[Dict[str, Any]] = None,
+        system_instruction: Optional[Any] = None,
     ) -> Any:
         """
         Async version of the generate content handler.
@@ -7472,6 +7476,7 @@ class BaseLLMHTTPHandler:
             contents=contents,
             tools=tools,
             generate_content_config_dict=generate_content_config_dict,
+            system_instruction=system_instruction,
         )
 
         if extra_body:

litellm/llms/gemini/google_genai/transformation.py

@@ -272,6 +272,7 @@ class GoogleGenAIConfig(BaseGoogleGenAIGenerateContentConfig, VertexLLM):
         contents: GenerateContentContentListUnionDict,
         tools: Optional[ToolConfigDict],
         generate_content_config_dict: Dict,
+        system_instruction: Optional[Any] = None,
     ) -> dict:
         from litellm.types.google_genai.main import (
             GenerateContentConfigDict,

tests/test_litellm/proxy/google_endpoints/test_google_api_endpoints.py

@@ -233,4 +233,80 @@ def test_google_stream_generate_content_with_cost_tracking_metadata():
         assert called_data["litellm_metadata"]["user_api_key_user_id"] == "test-user-id"
         assert called_data["litellm_metadata"]["user_api_key_team_id"] == "test-team-id"
         # Verify stream is set to True
-        assert called_data["stream"] is True
+        assert called_data["stream"] is True
+
+
+def test_google_generate_content_with_system_instruction():
+    """
+    Test that systemInstruction is correctly passed through from the endpoint to the router.
+    
+    This test verifies the fix for systemInstruction being dropped when forwarding
+    requests to Vertex AI through the Google GenAI endpoint.
+    """
+    try:
+        from fastapi import FastAPI
+        from fastapi.testclient import TestClient
+
+        from litellm.proxy.google_endpoints.endpoints import router as google_router
+    except ImportError as e:
+        pytest.skip(f"Skipping test due to missing dependency: {e}")
+    
+    # Create a FastAPI app and include the router
+    app = FastAPI()
+    app.include_router(google_router)
+    
+    # Create a test client
+    client = TestClient(app)
+    
+    # Mock all required proxy server dependencies
+    with patch("litellm.proxy.proxy_server.llm_router") as mock_router, \
+         patch("litellm.proxy.proxy_server.general_settings", {}), \
+         patch("litellm.proxy.proxy_server.proxy_config") as mock_proxy_config, \
+         patch("litellm.proxy.proxy_server.version", "1.0.0"), \
+         patch("litellm.proxy.litellm_pre_call_utils.add_litellm_data_to_request") as mock_add_data:
+        
+        mock_router.agenerate_content = AsyncMock(return_value={"test": "response"})
+        
+        # Mock add_litellm_data_to_request to pass through data unchanged
+        async def mock_add_litellm_data(data, request, user_api_key_dict, proxy_config, general_settings, version):
+            return data
+        
+        mock_add_data.side_effect = mock_add_litellm_data
+        
+        # Define the systemInstruction to test
+        system_instruction = {
+            "parts": [{"text": "Your name is Doodle."}]
+        }
+        
+        # Send a request with systemInstruction
+        response = client.post(
+            "/v1beta/models/gemini-2.5-pro:generateContent",
+            json={
+                "systemInstruction": system_instruction,
+                "contents": [
+                    {
+                        "parts": [{"text": "What is your name?"}],
+                        "role": "user"
+                    }
+                ]
+            },
+            headers={"Authorization": "Bearer sk-test-key"}
+        )
+        
+        # Verify the response
+        assert response.status_code == 200
+        
+        # Verify that agenerate_content was called
+        mock_router.agenerate_content.assert_called_once()
+        call_args = mock_router.agenerate_content.call_args
+        called_data = call_args[1]
+        
+        # Verify that systemInstruction is present in the call arguments
+        assert "systemInstruction" in called_data
+        assert called_data["systemInstruction"] == system_instruction
+        assert called_data["systemInstruction"]["parts"][0]["text"] == "Your name is Doodle."
+        
+        # Verify contents are also present
+        assert "contents" in called_data
+        assert len(called_data["contents"]) == 1
+        assert called_data["contents"][0]["role"] == "user"