Update app.py

app.py

@@ -10,14 +10,61 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
-class BasicAgent:
-    def __init__(self):
-        print("BasicAgent initialized.")
-    def __call__(self, question: str) -> str:
-        print(f"Agent received question (first 50 chars): {question[:50]}...")
-        fixed_answer = "This is a default answer."
-        print(f"Agent returning fixed answer: {fixed_answer}")
-        return fixed_answer
+import os
+import time
+from functools import lru_cache
+from openai import OpenAI, RateLimitError, APIError
+
+# --- GPT-4o-mini powered agent ------------------------------------------------
+class GPT4oMiniAgent:
+    """
+    Very small wrapper around OpenAI GPT-4o-mini.
+    • Reads the API key from the OPENAI_API_KEY env variable (set as a HF secret)
+    • Uses temperature-0 for deterministic grading
+    • Retries on transient rate-limit / 5xx errors with exponential back-off
+    • Caches identical questions in memory so reruns are near-instant
+    """
+    def __init__(self, max_retries:int=3, backoff:float=2.0):
+        api_key = os.getenv("OPENAI_API_KEY")
+        if not api_key:
+            raise EnvironmentError(
+                "OPENAI_API_KEY not found – add it in the Space secrets!"
+            )
+        self.client = OpenAI(api_key=api_key)
+        self.max_retries = max_retries
+        self.backoff = backoff
+        self.system_prompt = (
+            "You are a concise, precise assistant. "
+            "Answer fact questions directly. "
+            "When asked for code, return valid Python in fenced ``` blocks."
+        )
+        print("✅ GPT4oMiniAgent initialised.")
+
+    # in-memory cache so if the grader calls the same Q twice we don’t spend tokens
+    @lru_cache(maxsize=256)
+    def __call__(self, question: str) -> str:             # noqa: D401
+        print(f"🔹 Asking gpt-4o-mini: {question[:60]}…")
+        for attempt in range(1, self.max_retries + 1):
+            try:
+                response = self.client.chat.completions.create(
+                    model="gpt-4o-mini",             # ← official name :contentReference[oaicite:0]{index=0}
+                    messages=[
+                        {"role": "system", "content": self.system_prompt},
+                        {"role": "user", "content": question},
+                    ],
+                    max_tokens=512,
+                    temperature=0.0,
+                )
+                answer = response.choices[0].message.content.strip()
+                print(f"🔸 Answer (truncated): {answer[:60]}…")
+                return answer
+            except (RateLimitError, APIError) as e:
+                wait = self.backoff * attempt
+                print(f"⚠️  OpenAI error: {e}. Retry {attempt}/{self.max_retries} after {wait}s.")
+                time.sleep(wait)
+        # If all retries failed
+        return "I’m sorry, I could not generate an answer due to repeated API errors."
+
 
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """