Contributing

Project layout

This repository contains independent BigQuery and Snowflake dbt starter kits. Keep models, profiles, dependencies, and generated files inside the matching edition. Shared migration utilities live in migration/; common docs live in docs/.

Local development

Use Python 3.11 and the edition's hash-locked requirements. Run dbt from the directory containing that edition's dbt_project.yml; use its .ci profile for credential-free parsing. Check the edition README for profile setup before running a warehouse build.

Offline migration checks use synthetic fixtures and temporary local targets:

CI=true python -m unittest discover -s scripts/tests -v
CI=true python -m unittest discover -s migration/tests -v

Do not use production data in examples, tests, or pull requests. Never commit credentials, local profiles, .env files, Terraform variable files, dbt artifacts, or customer data. Keep direct identifiers out of curated marts.

Changes and review

Use two-space YAML indentation and lowercase snake_case SQL model names. Keep cleanup in staging models and business logic in marts. Add dbt schema tests for key constraints and relationships. Use concise imperative commit subjects, such as Add BigQuery customer value mart.

Pull requests should explain the user-facing change, identify the affected edition, and list the validation performed. A successful dbt parse checks project configuration; it does not prove warehouse SQL execution. Include screenshots for dashboard or other visual changes.