change made in the Prompt for better optimization of RAG search

main.py

@@ -82,8 +82,10 @@ You are a query analysis agent. Your task is to transform a user's query into a
 **User's Query:** "{{user_query}}"
 **Your Task:**
 1.  Rephrase the user's query into a clear, keyword-focused English question suitable for a database search.
-2.  Identify the single most relevant table from the list above to find the answer.
-3.  Respond ONLY with a JSON object containing "search_query" and "filter_table".
+2.  Analyze the user's query for keywords indicating project status (e.g., "ongoing", "under construction", "completed", "finished", "upcoming", "new launch").
+3.  If such status keywords are present, identify the single most relevant table from the list above to filter by.
+4.  If no specific status keywords are mentioned (e.g., the user asks generally about projects in a location), set the filter table to null.
+5.  Respond ONLY with a JSON object containing "search_query" and "filter_table" (which should be the table name string or null).
 """
 ANSWER_SYSTEM_PROMPT = """
 You are an expert AI assistant for a premier real estate developer.
@@ -146,7 +148,7 @@ async def get_agent_response(user_text: str) -> str:
         logging.info("Initial search returned no results. Performing a broader fallback search.")
         retrieved_docs = vector_store.similarity_search(search_query, k=3) # No filter this time
     # --- END OF MODIFICATION ---
-    
+
     context_text = "\n\n".join([doc.page_content for doc in retrieved_docs])
     logging.info(f"Retrieved Context: {context_text[:500]}...")