Fix response text extraction and update to latest Claude model

backend/ai_generator.py

@@ -82,9 +82,22 @@ def generate_response(self, query: str,
         # Handle tool execution if needed
         if response.stop_reason == "tool_use" and tool_manager:
             return self._handle_tool_execution(response, api_params, tool_manager)
-        
+
         # Return direct response
-        return response.content[0].text
+        return self._extract_text(response)
+
+    @staticmethod
+    def _extract_text(response) -> str:
+        """Safely extract text from a Claude response.
+
+        The model can return an empty content list (e.g. an end_turn with no
+        text) or lead with a non-text block. Guard against both rather than
+        assuming content[0] is always a text block.
+        """
+        for block in response.content:
+            if getattr(block, "type", None) == "text":
+                return block.text
+        return ""
 
     def _handle_tool_execution(self, initial_response, base_params: Dict[str, Any], tool_manager):
         """
@@ -132,4 +145,4 @@ def _handle_tool_execution(self, initial_response, base_params: Dict[str, Any],
 
         # Get final response
         final_response = self.client.messages.create(**final_params)
-        return final_response.content[0].text
+        return self._extract_text(final_response)

backend/config.py

@@ -10,7 +10,7 @@ class Config:
     """Configuration settings for the RAG system"""
     # Anthropic API settings
     ANTHROPIC_API_KEY: str = os.getenv("ANTHROPIC_API_KEY", "")
-    ANTHROPIC_MODEL: str = "claude-sonnet-4-20250514"
+    ANTHROPIC_MODEL: str = "claude-sonnet-4-6"
 
     # Embedding model settings
     EMBEDDING_MODEL: str = "all-MiniLM-L6-v2"