My dictionary sources seem to transliterate OCS, but not Greek. Since OCS has a number of weirder characters that most people who know Russian Cyrillic won't have come across, this sounds like a good model.

However, we should also be cognizant, for a "Latin/Greek/Sanskrit Vocabulary for Philologists", of what languages people will actually know. It's easier to see cognates between Greek and Avestan than between Greek and modern Persian, for example, but...how many of us actually know any Avestan? (While there are a larger number of people, even classicists, who know some basic Persian.) For the Latin spreadsheet, for example, it might be wise to have a "modern Romance descendents" box. And maybe German and English should be included in addition to Gothic, or Russian in addition to OCS (though the Slavic languages are close-knit enough that an OCS reflex is usually pretty transparent).

On the other hand, if we have too many boxes for cognates, we just increase clutter.

Do we have a standardized way to cite verbs? My intuition is to cite the 1s pres. ind. act. for Greek and Latin and the 3s for everything else. Doing otherwise gets unwieldy- in Old Irish you cite the 3s pres. ind. act, but in Sanskrit you generally cite the root followed by whatever forms are relevant, in many languages you just cite the infinitive, I don't even know what you do for Tocharian...

(Exceptions might reasonably be made, of course, for modern languages. The Spanish reflex of video is ver, not veo or ve, and to do otherwise is silly.)

Also, why does nobody care about Armenian?