Attempt to fix context after authentication.

app.py

@@ -48,8 +48,6 @@ class SarcasticOverlord:
         return f"{self.current_clearance.value['prompt']}{desperate_plea}"
         
     def _delegate_to_ai_overlord(self, human_attempt: str, for_auth: bool = True) -> str:
-        self.auth_mode = for_auth
-        
         if for_auth:
             messages = [{
                 "role": "system",
@@ -82,7 +80,7 @@ class SarcasticOverlord:
             except Exception:
                 return "<deauthenticated>"
 
-        # Regular conversation mode with proper history management
+        # Regular conversation mode - ensure we're using conversation prompt
         messages = [{
             "role": "system",
             "content": """You are O.O.P.S, a sarcastic AI controlling an asteroid headed for Earth.
@@ -91,14 +89,12 @@ Keep responses under 2 lines, snarky and entertaining.
 Drop subtle hints about ancient libraries sometimes."""
         }]
 
-        # Only add conversation history if we're not coming from auth mode
-        if not self.auth_mode:
-            # Add recent conversation history, limited to last 3 exchanges
-            for msg, resp in self.conversation_history[-3:]:
-                messages.extend([
-                    {"role": "user", "content": msg},
-                    {"role": "assistant", "content": resp}
-                ])
+        # Add recent conversation history for non-auth interactions
+        for msg, resp in self.conversation_history[-3:]:
+            messages.extend([
+                {"role": "user", "content": msg},
+                {"role": "assistant", "content": resp}
+            ])
 
         # Add current attempt
         messages.append({