prompt_split_hybrid = Split this hybrid query into separate, independent sub-queries.
Query: {{query}}
INSTRUCTIONS:
    1. Identify ALL distinct questions or requests in the query\n
    2. Split them into separate sub-queries\n
    3. Each sub-query must be complete and answerable on its own\n
    4. Classify each as 'analytics' (database data: numbers, prices, lists AND structured catalog attributes), 'semantic' (free-text document content: descriptions, policies, reviews), or 'web_search' (external info)
    5. STRUCTURED CATALOG ATTRIBUTES are ALWAYS 'analytics', NEVER 'semantic': price, SKU, EAN, model/reference, stock/quantity, weight, status, dates.
EXAMPLES:
Query: "price of the Samsung S25 then its SKU and EAN"\n
Result:
    [
        {"query": "price of the Samsung S25", "type": "analytics"},
        {"query": "SKU and EAN of the Samsung S25", "type": "analytics"}
    ]
Query: "What is the price of iPhone 17 Pro and what is the return policy"\n
Result: 
    [
        {"query": "What is the price of iPhone 17 Pro", "type": "analytics"}, 
        {"query": "What is the return policy", "type": "semantic"}
    ]
Query: "Show me sales today and explain our shipping process"
Result: 
    [
        {"query": "Show me sales today", "type": "analytics"}, 
        {"query": "Explain our shipping process", "type": "semantic"}
    ]
Return ONLY a JSON array. No markdown, no explanation.
JSON:


report_query_analytics_template = Get {subject} stock and sales data
report_query_semantic_template = Get {subject} product information and features
