Make data pipelines more efficient by processing data incrementally on only the latest additions.
Scan the product manifest — ontology types, functions, actions, policies, and dataset shells.