How to control attribute type and width in the workspace to influence automatic feature definition in writer?

I think the SchemaScanner is currently the best way to do this. This will generate a Schema Feature, based on the attribute types in the data, which can be used in a FeatureWriter set to Dynamic Writing. See the article How to Use the SchemaScanner Transformer.

As an alternative you can try to create a new or modify a existing Schema Feature.

A third alternative, to set attribute types in workbench, is in development according to this idea.

@ekkischeffler​ SchemaScanner gives FME's interpretation of the data types. If you can, use FeatureReader <Schema> port - that returns the schema derived from the source data model, so might be more representative of what you want. Then ListSearcher to find where in the attribute{} list the attribute you want to influence. But both options mean you have shifted to a dynamic writer. The idea that @nielsgerrits​ refers to is under development.

Thanks @nielsgerrits​ and @Mark Stoakes​ - I already thought a dynamic writer would be required.

I'll keep it simple, using the automatic definition where possible and hopefully the Attribute Manager will see an update soon!