Update app.py

app.py

@@ -25,7 +25,8 @@ class GeminiLLM(LLM):
     model_name: str = "gemini-2.0-flash"
     temperature: float = 0.1
 
-    
+
+
     @property
     def _llm_type(self) -> str:
         return "google-gemini-llm"
@@ -38,11 +39,11 @@ class GeminiLLM(LLM):
         }
         
         full_prompt = (
-            "You are an agent. Please respond concisely only with the answer, "
-            "no extra explanations!! stick to the question and use the tools.\n\n"
+            "You are a helpful agent that answers questions concisely and strictly follows instructions.\n"
+            "Respond ONLY with the requested information, no explanations or extra words. If the question specifies a format (number, name, comma separated list), follow it exactly.\n"
             f"Question: {prompt}"
         )
-    
+
         data = {
             "contents": [
                 {
@@ -240,6 +241,66 @@ def square_root(a: float) -> float:
         return a ** 0.5
     return sqrt(a)
 
+
+@tool
+def extract_numbers(text: str) -> List[float]:
+    """
+    Extract all numeric values from a given text.
+    Args:
+        text (str): Input text that may contain numbers.
+    Returns:
+        List[float]: A list of numbers found in the text.
+    """
+    import re
+    return [float(num) for num in re.findall(r'\d+(?:\.\d+)?', text)]
+
+@tool
+def extract_keywords(text: str, top_n: int = 5) -> List[str]:
+    """
+    Extracts the most frequent keywords from a text (ignores very common words).
+    Args:
+        text (str): The input text.
+        top_n (int): Number of keywords to return.
+    Returns:
+        List[str]: List of top keywords.
+    """
+    import re
+    from collections import Counter
+    stop_words = {"the", "a", "an", "and", "of", "in", "on", "for", "is", "at", "to", "by"}
+    words = re.findall(r'\b[a-zA-Z]+\b', text.lower())
+    filtered = [w for w in words if w not in stop_words]
+    return [word for word, _ in Counter(filtered).most_common(top_n)]
+@tool
+def extract_names(text: str) -> List[str]:
+    """
+    Extracts words that start with a capital letter (possible names or surnames).
+    Args:
+        text (str): The input text.
+    Returns:
+        List[str]: List of unique candidate names.
+    """
+    import re
+    names = re.findall(r'\b[A-Z][a-z]+\b', text)
+    return list(dict.fromkeys(names))
+
+@tool
+def find_non_commutative_pairs(table: Dict[str, Dict[str, str]]) -> List[tuple]:
+    """
+    Finds pairs (a,b) where the operation * is not commutative.
+    Args:
+        table (dict): A nested dictionary representing the operation table.
+    Returns:
+        List[tuple]: List of pairs where a*b != b*a.
+    """
+    non_commutative = []
+    elements = table.keys()
+    for a in elements:
+        for b in elements:
+            if table[a][b] != table[b][a]:
+                non_commutative.append((a, b))
+    return non_commutative
+
+
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 from langchain_core.prompts import PromptTemplate