dbt Wizard - First-Week Onboarding Tour

What you'll type, in order

Copy each prompt as written (recommended) or rephrase in your own words. Run the steps in order.

1. Project summary

   Start from the top, not from a model and not from a folder. dbt Wizard reads the project as a whole and returns the elevator pitch: what business this is, what the major subject areas are, and how the project is layered.

   Summarize what this dbt project does. What are the main subject areas and how is the project organized?

   Then group the models by layer and domain.

   List the staging, intermediate, and mart models. Group them by domain.

   Exercises: status, search, repo summarization
2. Inventory the models

   This is where the project goes from "a folder I just cloned" to "a map I can read." dbt Wizard returns the model list bucketed by layer and by domain, so you see both axes of the project on one screen.

   Show me the lineage, grain, and key columns for the orders mart model.

   Exercises: search, folder and tag grouping
3. Deep-dive on one mart

   We pick the orders mart because it sits at the busiest intersection of the project. If you can read this one model, you can read any of the others. Name the grain out loud - the question most new engineers skip and then regret.

   Show me a 10-row sample of the orders mart and the distinct values in the order_status column.

   Exercises: describe, lineage
4. Look at the actual data

   The distinct-values check on order_status is the load-bearing part of this step. Without it, Day-1 attendees reflexively write filters using values they made up. Write down the real distinct values; that's your reference list for the rest of the lab.

   What tests and contracts are defined on the orders model? Are any currently failing?

   Exercises: warehouse, dbt_show, select distinct
5. Tests and failure signal

   The test list tells you what the project's authors decided is true about this model. The list of currently failing tests is the single most useful piece of onboarding intel in the entire lab - that's where the project's pain points actually live.

   Create a new mart model called orders_by_week that aggregates orders to the week grain with order count, gross revenue, and distinct customers.

   Exercises: test introspection, run-results parsing
6. Ship something small

   This is the "ship something by Friday" deliverable. dbt Wizard writes orders_by_week.sql into models/marts/, built on the orders mart (not on staging - we respect the layering), with convention-aware naming and date truncation.

   Compile and preview orders_by_week. Don't materialize it.

   Exercises: file edits, convention-aware model generation
7. Compile and preview (no materialization)

   The model compiles, a sample of weekly rows renders, nothing lands in the warehouse. That's the safe-sandbox property Day 1 needs. The deliverable for this lab is the previewed, compiled, reviewable orders_by_week.sql file in the repo, not a built table.

   When the preview renders, confirm: roughly one row per week, populated non-zero values for order_count, gross_revenue, and distinct_customers, and a date column where every value is a week-start.
   Exercises: dbt_compile, dbt_show

Keep going