Growth Nirvana Bundle Query Transform

Growth Nirvana Bundle Query + Transform Router

Purpose

Use this as the high-level router skill. For best performance and consistency, delegate to one specialized skill based on user intent.

Decision Tree

• If the user asks analytical questions (channels, spend, CTR/CPC, table/field discovery), use the query workflow skill:
  • ../growth-nirvana-bundle-query/SKILL.md
• If the user asks to draft/edit transformation SQL, use the authoring workflow skill:
  • ../growth-nirvana-transform-authoring/SKILL.md
• If the request mixes both, run query workflow first (to select model + fields), then authoring workflow.

Shared Constraints

• Always resolve dataset by dataset/dataset.json.displayName.
• Always perform model selection before querying or drafting.
• Never assume metric names (_cost vs cost) without checking field metadata.
• Prefer bundle evidence; use MCP for validation/execution.
• Include result rows only when requested (include=results, row_limit).

Response Contract

Always include:

• workflow: bundle-query-transform@v2
• dataset display name + dataset id
• selected model/table + rationale
• action performed (query or authoring)
• evidence paths
• warnings/conflicts and next steps

Additional Resources

• Query workflow skill: ../growth-nirvana-bundle-query/SKILL.md
• Authoring workflow skill: ../growth-nirvana-transform-authoring/SKILL.md
• Existing deep reference docs in this folder:
  • reference.md
  • examples.md
  • common-queries.md