Yeah, as I said I was commenting more on the UX. I do think it has got tremendous potential and I'm not berating it by any means. Just saying that in terms of polish I don't think I'd want to be working with it 8 to 16 hours a day every day in production as it currently stands.
As for the runtime material publishing, all I can see in the docs is:
When exporting for the Godot game engine, Material Maker will generate a .tres file that describes a fully configured SpatialMaterial.
Which doesn't sound like what I'd want. And even if they mean ShaderMaterial I have my doubts, would be better tho.
Honestly the same goes for godot's current state of VisualShaders. I think long term it has a bright future, but it's not currently very nice to work with and I'd rather write shaders in code as it stands. But given some additional major features such as compute shaders and most importantly lots of QoL polish, such as node groups and an easy way to preview any nodes output on the viewport mesh(a la node wrangler and viewer nodes in blender) or frames for commenting sections of your graph I think it'll be great.