mongofiles can currently only get files from GridFS by filename.
However filenames are not unique on GridFS, so if multiple files have the same name on GridFS, then mongofiles gets one of them, in an undefined behavior.
get by _id would be a great feature.
Currently I have to patch the DB to make the filename unique for the duration of the download, which is not great as it breaks things during that time.
A full query could also be useful, but not strictly needed: the query could be done with the mongo shell, and then downloads by _id with mongofiles.