How dbt-labs/jaffle-shop works
Project type and purpose
• This is a data-transformation (“ELT”) project built with dbt (data build tool).
• Its purpose is to ingest raw CSV data (“jaffle-data” or S3) into a warehouse, apply staging transformations, and build presentation-ready marts.
• It’s an open-source learning project (the “Jaffle Shop”) demonstrating dbt Cloud workflows, CI/CD, environments, and dbt model organization.
File structure
• Top-level config:
• dbt_project.yml (defines project name, model folders, seed paths)
• packages.yml (external dbt packages)
• Taskfile.yml (automation tasks)
• GitHub workflows under .github/workflows (CI, CD for prod/staging)
• Data sources:
• jaffle-data/ (example CSV seeds)
• seeds/ (empty by default, can be used for controlled reference data)
• Macros: reusable SQL snippets in macros/
• Models:
• models/staging/ (one-to-one raw → stg_ tables)
• models/marts/ (business-centric tables, metrics, time spine)
• Tests: data-tests/ and analyses/ hold tests and exploratory queries
• CI/CD scripts: workflows trigger dbt Cloud runs or CLI commands via dbt Cloud CLI scripts.
README for additional insights
• Describes two environments: “staging” branch and “main” branch tied to respective dbt Cloud environments/schemas.
• Outlines how data is loaded (dbt seed for CSV, S3 copy, or jafgen tool).
• Highlights UI (dbt Cloud IDE) vs. CLI workflows.
• Mentions lineage (DAG) exploration in dbt Cloud Explorer.