Well, so in order to share what we got the best, I'm offering a dataset called "Inner Sun". It won't be all about agartha. But they are not tagged, but properly captioned via civitai.
In short: 33/34 images, captions (descriptions),as just tags seem to be subpar here.
Even SDXL can deal with caption. Pony and Illustrious do not. Flux and similar work fine with such captions.
Good thing about captions is that they point out elements discussed on the Agartha… See the full description on the dataset page:
https://huggingface.co/datasets/semiosphere/innersun_dataset.
