What problem does it solve?
The dive skill eliminates ad-hoc dashboard work by enforcing a grain-first, repeatable workflow that produces a dbt marts model, a manifest-driven data contract, a React TSX dive component, DivesPage registration, and a set of verification seed rows so dashboards are correct, auditable, and maintainable.
Core Features & Use Cases
- Three Questions gating: forces question, audience, and visualization clarity before any SQL is written.
- Grain-first model design: guided grain definition, dimension listing, and filter compatibility to avoid ratio and aggregation bugs.
- End-to-end artifact creation: produces dbt marts SQL, manifest file, TSX component wired to useDiveData/useDiveVerifications, DivesPage registration, and adds ~15–20 rows to the shared verifications seed.
- Iterative dev loop & testing: dbt run/seed/test guidance, vite-dbt manifest sync notes, and artifact output for verification and promotion.
- Use case: building a new corporate dashboard (e.g., MA enrollment by company) with verify checks and methodology content embedded for auditability.
Quick Start
Build a new dive for "MA enrollment by company" and guide me through the three questions, grain design, dbt marts SQL, manifest, TSX component, and verify checks.