Update data_planner.py

data_planner.py

@@ -29,6 +29,13 @@ def determine_data_requirements(
             f"Current participantId: '{participant_id or ''}'\n"
             "Use this schema for all reasoning. It includes field names, types, and aliases:\n"
             f"{json.dumps(schema)}\n\n"
+            "PLANNING RULES:\n"
+            "- Choose collections ONLY if they contain the fields needed to answer.\n"
+            "- If the user asks about headcount/revenue history, choose 'participants' (fields: headcountHistory, revenueHistory).\n"
+            "- Apply companyCode ONLY on collections that have companyCode in the schema (do NOT add companyCode to 'participants').\n"
+            "- If an alias is used in the question, map it to the canonical field using $aliases.\n"
+            "- Return 'collections' as a list of objects [{'name': <collection>, 'fields': [<field1>, ...]}].\n"
+            "- 'filters' must only include fields that exist in the chosen collections.\n"
             "Strictly return only a valid JSON object with keys: collections, filters, instruction. No explanation."
         ),
     }