I’ve worked around this sort of thing with loops.
Check the output with a code node → feed result to an if node → feedback to list with some params
It’s not necessarily an less annoying. But maybe worth it if you have to do this frequently.
I imagine there will some day be a better way to handle some of that gymnastics.
Oh, another hack that this comment brings to mind.
You can do both!
A data explorer node with “raw” sql query might be able to get past the 100 topic limit, especially if you have a little filtering in there. AI is helpful for authoring queries.
So you could encapsulate a little bit of the logic in sql still, while leaving the rest in a form that’s easier for you to understand and modify over time.