Growth Nirvana Bundle Query

Growth Nirvana Bundle Query

Purpose

Answer questions from bundle files first, then optionally run live queries via MCP.

Workflow

1. Resolve dataset by matching dataset/dataset.json.displayName.
2. Select reporting model:
   • inspect models/combined_report* and related downstream models
   • choose combined_report or variant (combined_report_ff, combined_report_catalog, etc.)
3. Validate fields in:
   • warehouse/tables/<selected_table>/fields/*.json
4. Build SQL based on validated fields.
5. If execution is needed:
   • run create_query_execution
   • fetch with get_query_execution(include=results, row_limit=...)
6. Return answer with evidence and rationale.

Guardrails

• Do not assume cost; verify _cost/cost.
• If variant logic changes semantics (for example Meta/Bing/X remaps), state that clearly.
• If no rows returned, provide date-coverage diagnostic query.

Response Contract

• workflow: bundle-query@v1
• dataset and selected model/table
• answer summary
• SQL used (or summary)
• execution id/run state when executed
• result rows summary and truncation flag when requested
• evidence paths and caveats

Resources

• Query templates: common-queries.md
• Usage examples: examples.md
• Troubleshooting: troubleshooting.md