Extend Missing Primary Key Tests to include dbt.constraints PK test

[Pirognoe]

Describe the feature

I would like to add additional option no. 4 to Missing Primary Key Tests test to include dbt_constraints.primary_key test from https://github.com/Snowflake-Labs/dbt_constraints/tree/main

Describe alternatives you've considered

A clear and concise description of any alternative solutions or features you've considered.

Additional context

After adding a snippet like below

columns:
      - name: account_sid
        description: Primary key
        data_tests:
          - dbt_utils.at_least_one
          - dbt_constraints.primary_key

I already check my column for PK validity, but for Missing Primary Key Tests I still need to add the following test, which are essentially duplicates of the above

constraints:
          - type: not_null
        data_tests:
          - dbt_constraints.primary_key
          - dbt_utils.at_least_one
          - unique

Who will this benefit?

Teams using both packages

Are you interested in contributing this feature?

[b-per]

Can you overwrite the default combination of tests we look at to raise missing PK? https://dbt-labs.github.io/dbt-project-evaluator/latest/customization/overriding-variables/#testing-and-documentation-variables
(just be aware that this var is a list of list (e.g. a list of combination of tests)

I think that having would work and not report error on any table where a columns has dbt_constraints.primary_key set.

[["dbt_constraints.primary_key"], ...]

[github-actions]

This issue has been marked as Stale because it has been open for 180 days with no activity. If you would like the issue to remain open, please comment on the issue or else it will be closed in 7 days.

[migueldichoso]

We have a customer who raised a feedback related to improving the logic behind the Missing Primary Key Tests. Based on Testing > Missing Primary Key Tests document, this tests will list any model that does not have either: a not_null test AND a unique test applied to a single column, or, a dbt_utils.unique_combination_of_columns test applied to a set of columns.

However, from their POV, re-introducing not_null data tests on primary key columns feels like a step backwards, as these constraints are already enforced at the database level.

Key points:

• Data tests add overhead to job runs.
• These tests only run after models are built, so they don’t stop nulls or duplicates from being inserted
• If a column is already defined with a NOT NULL constraint, a data test offers no additional protection or insight.

Suggestion: Promote the use of contracts and constraints as best practices for these cases. Since they’re enforced by the database itself, they offer real-time protection—unlike post-run tests. Ideally, the recommendation system should check for existing constraints and suggest them as a remediation strategy alongside (or instead of) data tests.