Skip to content

Model-Generation.md

Latest commit

 

History

History
50 lines (35 loc) · 1.84 KB

Model-Generation.md

File metadata and controls

50 lines (35 loc) · 1.84 KB
Raw

Model Generation

/generate-model turns the proof-backed refactor, profile, and approved test spec into dbt artifacts.

Prerequisites

  • manifest.json
  • dbt/ scaffolded by ad-migration setup-target
  • completed scoping
  • completed profiling
  • approved test spec
  • completed refactor state for the object

Invocation

/generate-model silver.DimCustomer silver.FactInternetSales

Pipeline

  1. generate dbt SQL and schema YAML
  2. run dbt compile against runtime.target
  3. materialize direct parents with an empty run when unit tests need them
  4. run scoped dbt unit tests for the generated model
  5. self-correct up to the command limits
  6. run the independent model review loop
  7. commit successful items and optionally raise a PR

The generator uses the proof-backed refactor persisted by /refactor-query; it does not go back to raw procedure SQL as the primary migration input.

/generate-model does not run a broad dbt build for each generated mart model. Target setup uses dbt build for generated staging/source setup artifacts, and /refactor-mart uses candidate-scoped validation from its approved plan.

What gets written

  • model SQL under dbt/models/
  • schema YAML with tests and unit_tests:
  • snapshot artifacts when the target pattern requires them

Notes

  • Successful items are committed as they finish.
  • Items marked is_source: true are skipped because they are not migration targets.
  • /status is the best way to see which objects are now ready versus still blocked.
  • runtime.target and runtime.sandbox are both required: target for dbt validation, sandbox for any live source-backed checks the workflow performs.
  • See [[Target Setup]] for the source-facing dbt validation created before model generation.

Next step

Use [[Status Dashboard]] to plan the next batch or merge the PR opened by the run.