Update app.py

app.py

@@ -19,6 +19,7 @@ from agents.extensions.models.litellm_model import LitellmModel
 
 #os.getenv("OPENAI_API_KEY")
 os.getenv("GEMINI_API_KEY")
+os.getenv("TAVILY_API_KEY")
 
 
 # add this
@@ -27,6 +28,29 @@ nest_asyncio.apply()
 
 #Tools
 
+@function_tool
+def tavily_search(query: str) -> str:
+    """
+    Perform a Tavily web search.
+    Args:
+        query (str): The search query string.
+    Returns:
+        str: Formatted search results.
+    """
+    try:
+        client = TavilyClient()
+        results = client.search(query=query, max_results=5)
+
+        formatted = []
+        for result in results.get("results", []):
+            formatted.append(f"**Title**: {result['title']}\n**URL**: {result['url']}\n**Content**: {result['content']}\n")
+
+        return "\n\n".join(formatted) or "No results found."
+
+    except Exception as e:
+        return f"Error using Tavily Search: {e}"
+
+
 @function_tool
 def web_search(query: str) -> str:
     """
@@ -115,7 +139,7 @@ class BasicAgent:
         instructions = """
 You are a ReAct (Reason-Act-Observe) agent that searches the internet to find accurate answers to questions.
 ## Available Tools
-- **web_search**: Search the web for information
+- **tavily_search**: Search the web for information
 - **visit_website**: Visit specific webpages for detailed content
 ## Output Format Rules
 Your final answer must be **exactly one** of these formats:
@@ -155,7 +179,7 @@ When you've found the answer, stop reasoning and return ONLY the answer.
             name="Expert Question Answering Agent",
             instructions=instructions,
             tools = [
-                web_search,
+                tavily_search,
                 visit_website
             ],
             model=LitellmModel(
@@ -170,8 +194,12 @@ When you've found the answer, stop reasoning and return ONLY the answer.
                 input=question,
                 max_turns=25
             )
-
-        
+        print("\n--- Intermediate Reasoning ---")
+        for step in result.steps:
+            print("🧠 Thought:", step.thought)
+            print("⚙️ Action:", step.tool_call)
+            print("🔍 Observation:", step.observation)
+            print("-" * 50)
 
         print(f"Agent returning fixed answer(first 50 chars): {result.final_output[:50]}...")
         return result.final_output