Revert app.py to original BasicAgent design for regression testing

app.py

@@ -23,102 +23,13 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 # --- Basic Agent Definition ---
 class BasicAgent:
-    """A LlamaIndex-based agent."""
     def __init__(self):
         print("BasicAgent initialized.")
-        # Initialize the core components
-        self.llm = self._initialize_llm()
-        
-        # Import get_tools from agent.py here to avoid circular imports
-        from agent import get_tools
-        self.tools = get_tools()
-        
-        self.memory = ChatMemoryBuffer.from_defaults(token_limit=3900)
-        # Build the agent
-        self.agent = self._build_agent()
-        print("Agent setup complete.")
-        
-    def _initialize_llm(self) -> LLM:
-        """Initialize the LLM based on configuration."""
-        provider = os.getenv("DEFAULT_LLM_PROVIDER", "gemini").lower()
-        
-        if provider == "gemini":
-            api_key = os.getenv("GOOGLE_API_KEY")
-            if not api_key:
-                raise ValueError("GOOGLE_API_KEY not found in environment variables")
-            
-            return Gemini(
-                model_name="models/gemini-1.5-flash",
-                api_key=api_key,
-                temperature=0.1,
-                top_p=0.95,
-                max_tokens=1024,
-            )
-            
-        elif provider == "huggingface":
-            api_key = os.getenv("HUGGINGFACE_API_KEY")
-            if not api_key:
-                raise ValueError("HUGGINGFACE_API_KEY not found in environment variables")
-            
-            return HuggingFaceInferenceAPI(
-                model_name="Qwen/Qwen2.5-Coder-32B-Instruct",
-                api_key=api_key,
-                temperature=0.1,
-                max_tokens=1024,
-            )
-        else:
-            raise ValueError(f"Unsupported LLM provider: {provider}")
-    
-    def _build_agent(self) -> ReActAgent:
-        """Build and return the agent."""
-        # Load system prompt from file and append output format requirements
-        try:
-            with open("system_prompt.txt", "r", encoding="utf-8") as f:
-                system_prompt = f.read()
-            # Append output format to system prompt
-            system_prompt = f"{system_prompt}\n\nIMPORTANT OUTPUT FORMAT:\n{OUTPUT_FORMAT}"
-        except Exception as e:
-            print(f"Error loading system prompt: {e}")
-            system_prompt = f"You are an intelligent agent designed to answer a wide variety of questions.\n\nIMPORTANT OUTPUT FORMAT:\n{OUTPUT_FORMAT}"
-        
-        return ReActAgent.from_tools(
-            tools=self.tools,
-            llm=self.llm,
-            memory=self.memory,
-            system_prompt=system_prompt,
-            verbose=True,
-        )
-
     def __call__(self, question: str) -> str:
         print(f"Agent received question (first 50 chars): {question[:50]}...")
-        try:
-            # Process the question
-            response = self.agent.query(question)
-            answer_text = str(response)
-            
-            # Extract the FINAL ANSWER part if it exists
-            if "FINAL ANSWER:" in answer_text:
-                reasoning_trace = answer_text.split("FINAL ANSWER:")[0].strip()
-                model_answer = answer_text.split("FINAL ANSWER:")[1].strip()
-                
-                # Include the reasoning trace in the response but formatted for JSON
-                result = {
-                    "model_answer": model_answer,
-                    "reasoning_trace": reasoning_trace
-                }
-                
-                # Return just the answer part for direct evaluation
-                print(f"Agent generated answer: {model_answer[:50]}..." if len(model_answer) > 50 else f"Agent generated answer: {model_answer}")
-                return json.dumps(result)
-            else:
-                # If no FINAL ANSWER pattern, return the whole response
-                print(f"No 'FINAL ANSWER' found in response. Returning full response.")
-                return json.dumps({"model_answer": answer_text, "reasoning_trace": ""})
-                
-        except Exception as e:
-            print(f"Error generating answer: {e}")
-            error_msg = f"I encountered an error while answering your question: {str(e)}"
-            return json.dumps({"model_answer": error_msg, "reasoning_trace": ""})
+        fixed_answer = "This is a default answer."
+        print(f"Agent returning fixed answer: {fixed_answer}")
+        return fixed_answer
 
 def run_and_submit_all(profile: gr.OAuthProfile | None):
     """