More broadly, how we cite data hasn't caught up to the scale of these datasets. Separately, I found that genome assembly databases (like NCBI) often don't have originating publications linked for a genome. Sometimes there is no paper, but many times there is a paper, it is just missing.

What I can do now is to keep working on recovering the originating publication for most of these assemblies from NCBI and Europe PMC, and improve table S1 on Zenodo and the relevant github repos published in this paper (chrombase, genbargo, egt).

That's fair. We weren't able to cite thousands of papers in the manuscript, but table S1 lists every accession and its NCBI submitter. In the manuscript we acknowledge institutions at the >20-genome level, but that under-credits the many groups contributing fewer genomes.