Update app.py

app.py

@@ -17,7 +17,6 @@ MODEL_NAME = "deepseek-ai/Deepseek-V3"
 class BasicAgent:
     def __init__(self):
         print("BasicAgent initialized.")
-        load_dotenv()
         self.api_key = os.getenv("TEST_AGENT_KEY")
         if not self.api_key:
             raise ValueError("OpenRouter API Key not found.")
@@ -47,42 +46,41 @@ class BasicAgent:
         print(f"Agent tools initialized: {list(self.tools.keys())}")
 
     def _call_llm(self, conversation_history: list) -> str:
-        print(f"Calling LLM. Conversation history length: {len(conversation_history)}")
-        headers = {
-            "Authorization": f"Bearer {self.api_key}",
-            "Content-Type": "application/json",
-            "HTTP-Referer": os.getenv("SPACE_ID", "http://localhost"), # Recommended by OpenRouter
-            "X-Title": os.getenv("SPACE_TITLE", "Test Agent")      # Recommended by OpenRouter
+    print(f"Calling LLM. Conversation history length: {len(conversation_history)}")
+
+    # Hugging Face headers (simpler)
+    headers = {
+        "Authorization": f"Bearer {self.api_key}",  # should be your HF token
+        "Content-Type": "application/json"
+    }
+
+    # Format conversation into a single prompt string
+    # You can improve this formatting later if needed
+    prompt = ""
+    for turn in conversation_history:
+        role = turn.get("role", "user").capitalize()
+        content = turn.get("content", "")
+        prompt += f"{role}: {content}\n"
+    prompt += "Assistant:"
+
+    # Payload for Hugging Face
+    payload = {
+        "inputs": prompt,
+        "parameters": {
+            "temperature": 0.7,
+            "max_new_tokens": 512
         }
-        payload = {
-            "model": self.model_name,
-            "messages": conversation_history,
-            "temperature": 0.7, # Adjust as needed
-            # "max_tokens": 1000 # Adjust as needed
-        }
-        try:
-            response = requests.post(self.llm_api_url, headers=headers, json=payload, timeout=120)
-            response.raise_for_status()
-            llm_response_data = response.json()
-            if llm_response_data.get("choices") and llm_response_data["choices"][0].get("message"):
-                content = llm_response_data["choices"][0]["message"].get("content", "").strip()
-                print(f"LLM raw response: {content[:200]}...")
-                return content
-            else:
-                print(f"LLM response malformed: {llm_response_data}")
-                return "Error: LLM response was malformed."
-        except requests.exceptions.Timeout:
-            print("Error: LLM API call timed out.")
-            return "Error: LLM call timed out."
-        except requests.exceptions.RequestException as e:
-            print(f"Error calling LLM API: {e}")
-            if e.response is not None:
-                print(f"LLM Error Response Status: {e.response.status_code}")
-                print(f"LLM Error Response Body: {e.response.text}")
-            return f"Error: Failed to communicate with LLM. {str(e)}"
-        except Exception as e:
-            print(f"An unexpected error occurred during LLM call: {e}")
-            return f"Error: An unexpected error occurred communicating with LLM. {str(e)}"
+    }
+
+    url = "https://api-inference.huggingface.co/models/deepseek-ai/DeepSeek-V3"
+    response = requests.post(url, headers=headers, json=payload)
+
+    try:
+        return response.json()[0]["generated_text"].split("Assistant:")[-1].strip()
+    except Exception:
+        return f"Error: {response.text}"
+
+        
 
     def __call__(self, question_data: dict) -> str:
         task_id = question_data.get("task_id")