It makes impossible to use the workflow for multiple sheets. In my case, the sheetId is dynamic and passed as an argument into the workflow, so the same workflow is used to write rows in different sheets with different schemas.
I expected that “Map Automatically” means that it reads the columns of the sheet during runtime, but as I see now, it caches them during build time and tries to reuse whatever cached schema independently of the actual sheet referenced by sheetId.
Hope it will be fixed soon! thanks!