To be specific, each stage of the setup pipeline will consist of a small component that consumes and produces a file or object (or group of files/objects) that specifies:
the topology (atomic elements, connectivity, and likely bond orders) of polymers, small molecules, ions, etc.
the atomic positions for some or all of the atoms
potentially, parameters assigned to some or all of the molecules/residues
One of the first things that should be decided upon, is the kind of input and output file formats that will eventually be supported here.
A related discussion we should probably take into account can be found on alchemsitry.org github page: https://github.com/alchemistry/fileformat