feat: Enhance generation configuration by adding system instruction to Vertex AI model requests

cogs/aimod.py

@@ -1095,17 +1095,22 @@ CRITICAL: Do NOT output anything other than the required JSON response.
 
         try:
             print(f"Querying Vertex AI model {model_path}...")
-            # Use the model's system_instruction parameter if available and preferred
-            # For now, the system prompt is embedded in the user_context_text
-            model_instance = self.genai_client.get_generative_model(
-                model_name=model_path,
-                system_instruction=types.Content(role="system", parts=[types.Part(text=system_prompt_text)]) # Pass system prompt here
+
+            # Prepare the generation config with system instruction
+            # The existing 'generation_config' (lines 1063-1072) already has temperature, max_tokens, safety_settings.
+            # We need to add system_instruction to it.
+            final_generation_config = types.GenerateContentConfig(
+                temperature=generation_config.temperature, # from existing config
+                max_output_tokens=generation_config.max_output_tokens, # from existing config
+                safety_settings=generation_config.safety_settings, # from existing config
+                system_instruction=types.Content(role="system", parts=[types.Part(text=system_prompt_text)])
+                # response_mime_type="application/json", # Consider if model supports this for forcing JSON
             )
 
-            response = await model_instance.generate_content_async(
-                contents=request_contents, # This will now only contain the user turn with context and images
-                generation_config=generation_config,
-                # stream=False # Default
+            response = await self.genai_client.aio.models.generate_content(
+                model=model_path, # Correctly formatted model path
+                contents=request_contents, # User's message with context and images
+                generation_config=final_generation_config, # Pass the config with system_instruction
             )
             
             ai_response_content = self._get_response_text(response)