I don't know exactly how far back the USDA MyPlate goes, but in their 2014 evidence review there were at least a handful of trials that informed their conclusions, though certainly the vast majority were prospective cohort studies.

More of both types of studies were then captured in their 2020 evidence review, but the majority were still observational there as well.

I get what you're saying, the number of RCTs incorporated into USDA recommendations is definitely small, but it's not quite as small as zero.