So the after pipeline run hook has access to the catalog as a live object so you should be able to access all of the SQL dataset instances