You could do it, but the programming costs are most probably prohibitive. If you already have the model solution in non-Dynare form I would recommend that you use the state-space form to generate the model moments yourself. You can basically find the formulas in Kim, Kim, Schaumburg. Sims.
Jpfeifer, I still have not solved this problem. Could you help explain what you mean by use the state form to generate the model moments? I read the paper that you suggested and I think the equations of interest are (39) -(42) but I do not know how to use them to generate the moments.