Hey @maria! Great question!
Since we have control over seeds, we’ll make sure that the data is already in a “staging” format, and name it with the stg_ prefix. Since there isn’t usually a data source for these models (typically they are codified business logic), we’ll often end up with names like stg_country_codes or stg_email_exclusion_list, rather than following a stg_<source>_<object> format. We aren’t super strict on this convention though, so we’re open to feedback here!
This is kind of hacky, but you can actually document and test seeds in a .yml files in the models/ directory. We’ll improve this in a future version of dbt, but for now you can do:
version: 2
models:
- name: stg_country_codes
columns:
- name: country_code
tests:
- unique
- not_null