It's not clear what you are looking for (alternatively, your expectation is unrealistic). Drawing on the Cree example, you can locate the gloss "2-like-DIR?-N thus:CNJ-speak.Cree-3" or "1-see-INV-3" in that database. I assume you did not know that you have to enter those particular glosses (why not "see-1-3-INV"? – because the gloss reflects the morphemes of Cree, their order, and their conventional abbreviation in some cases like INV). It seems really unlikely that you would ever want to look up a gloss like "1-see-INV-3", unless you are doing a comparative study of Algonkian. You can't compare sentence structures across languages in any meaningful way, if you are searching by gloss. But if you were searching using English translations, you might have somewhat better odds of learning something. For example, you might get the equivalent of "You see me" from Cree and some other language (if it's in the database). It is not in the Tibetan data, nor the Arabic.
You might contact Giellatekno. There's a reasonable chance that they have large amounts of text, much of which may be already parsed, and a gloss plus translation is in principle available, at least for the Saami languages. You would probably have to build the database yourself, but they have good computational tools for parsing, and the main problem would be associating lexical items with something that you might use (presumably you don't want "goahti" in your gloss, you want the English equivalent).