after_catalog_created hook with after_context_created to get the context. I did the same With before pipeline run you won't be able to access those datasets with say kedro ipython. Therefore there's no point in saving them But if you use my way you'll have to namespace your datasets uniquely