At the moment, there is no easy way to exclude computing coverage for models found within deeply nested file structures.

e.g. take for example a dbt project with the following file structure

📂models 
    📂data_mart
        - dim_customers.sql
        📂bronze 
          - data_mart__bronze_cleaning_model.sql
        📂silver
          - data_mart__silver_transforming_model.sql
          - data_mart__silver_transforming_model_2.sql

         
   📂data_mart_2
        - dim_model_1.sql
        - fct_model_2.sql
        📂bronze
          - data_mart_2__bronze_cleaning_model.sql
        📂silver
          - data_mart_2__silver_transforming_model.sql

If you only wanted to perform docs coverage for the final tables 'gold' medallion tables (e.g. /models/data_mart/dim_cusomters.sql, /models/data_mart_2/dim_model_1.sql and /models/data_mart_2/fct_model_2.sql), there is no way to do that currently.

We should have a feature available where we can exclude models from being computed in order to reduce the amount of noise that occurs during the coverage comparison report. This feature should work with the existing --model-path-filter argument, which allows for more finer grain control of what to 'include', but not what to 'exclude'