>>67645it uses a document-based database to store tags and metadata.
to add new stuff you can either drag and drop the things you want to add into the gui or if they already have record files made for them you can just mass dump them where ever you want inside whatever folder you designated to be your library.
the gui looks and works fairly similar to a booru. there's a search bar you can put your query into, currently the search system is strong enough to do a few different types of searches and handles a few different types of tags. after you submit a search, multithreading is used to rapidly dig through your libraries and display results as they're being found in real time.
what makes it less intrusive than hydrus, is that filebrary supports multiple libraries. it also doesn't rename your files into hash garbage. it doesn't use sqlite or mongodb, all the database records are indiviual files that are human-readable in toml markup.