Hi Gustavo,
You're right, the lib files are for modifying Amber so that you can simulate glycoconjugates.
I looked into MDWeb a bit more. They would have to add GLYCAM as an option. However, MDWeb looks to implement AMBER99SB using either Gromacs or NAMD, so adding GLYCAM will come with the usual problems: AMBER scales it's 1-4 interactions and both Gromacs and NAMD handle that, but GLYCAM does not scale it's 1-4 interactions... So to use both GLYCAM and AMBER parameters you need "mixed" scaling. Neither NAMD nor Gromacs can do that afaik and the effects of not doing this will be system specific.
Marko Wehle of the "Max Planck Institute of Colloids and Interfaces" created a script to properly convert from Amber/Glycam topologies to Gromacs. Perhaps you could contact WebMD about creating and using your own topology files. I can forward you Marko's script if that turns out to be possible.
On top of all those problems, I must warn you that converting from "native" software comes with a lot of extra complexity. It's very easy to make mistakes. If it's possible for you to get AMBER ($500 I believe) and access to computing resources at your institution, that may well turn out easier in the long run.
Good luck,