Right now the feature seems to be public available. It is possible to schedule jobs with ADLS passthough enabled and do not have to provide service principal credentials.

However I ask myself how that works behind the scenses. When working interactively with notebooks the "passthrough" refers to the fact that the user is logged in and already available access_token can be used for access ADLS. Job clusters runs unattended and there is no access_token available...

Does anybody know which OAuth2 flow Databricks use in order to get the access_token of the job owner?