Packages — dbt_utils and the Ecosystem

Install

packages.yml:

packages:
  - package: dbt-labs/dbt_utils
    version: 1.3.0
  - package: calogica/dbt_expectations
    version: 0.10.4
  - package: dbt-labs/codegen
    version: 0.13.1

Run dbt deps. Packages land in dbt_packages/ (gitignored).

dbt_utils — mandatory

The first package everyone installs.

Surrogate keys

{{ dbt_utils.generate_surrogate_key(['customer_id', 'effective_from']) }}

Star

SELECT {{ dbt_utils.star(from=ref('stg_orders'), except=['_fivetran_synced']) }}
FROM {{ ref('stg_orders') }}

Pivot

SELECT user_id, {{ dbt_utils.pivot('order_status', ['pending', 'paid', 'cancelled']) }}
FROM {{ ref('fct_orders') }} GROUP BY 1

Recency test

tests:
  - dbt_utils.recency:
      datepart: day
      field: order_placed_at
      interval: 1

Expression is true

- dbt_utils.expression_is_true:
    expression: "row_count_today > 0.5 * row_count_yesterday"

Date spine

{{ dbt_utils.date_spine(
    datepart="day",
    start_date="cast('2024-01-01' as date)",
    end_date="cast(current_date as date)"
) }}

dbt_expectations

Expressive tests inspired by Great Expectations:

columns:
  - name: revenue
    tests:
      - dbt_expectations.expect_column_values_to_be_between:
          min_value: 0
          max_value: 1000000

codegen

Generates boilerplate:

dbt run-operation generate_source --args 'schema_name: raw_shopify'
dbt run-operation generate_base_model --args '{source_name: shopify, table_name: orders}'

Saves weeks of typing.

dbt_artifacts

Captures dbt run history to your warehouse. Useful for slow-model monitoring and SLA dashboards.

elementary

Open-source data observability. Anomaly detection on top of dbt tests.

Source-system packages

fivetran/shopify, fivetran/stripe, etc. — pre-built staging models. If your source is on Fivetran, there's probably a package.

Versioning

- package: dbt-labs/dbt_utils
  version: 1.3.0           # exact pin in prod

Common mistakes

1. Not using packages — reinventing dbt_utils.
2. Pinning too loosely in prod.
3. Modifying installed package code — never edit dbt_packages/.

Takeaway

dbt_utils mandatory. dbt_expectations for richer tests. codegen for boilerplate. Pin exact in production.