If you don't expect the solid to change which table it's writing to once it's used in a pipeline (and so you're just writing it this way so that it can be reused in different pipelines for different tables), you might want to try out the solid factory pattern (https://docs.dagster.io/concepts/solids-pipelines/solids#solid-factory), which would allow you to "bake in" the metadata for that output, instead of getting it from the inputs. However, that may not work for you depending on what logic is actually going on in the solid

The other option would be to bundle more information into the output itself by creating some sort of wrapper object (maybe a namedtuple that contains the data you're currently putting in the output + fields target_schema/target_table) and returning that instead. This would give the IOManager enough information to figure out where to put that data