For Weiner et al. (1984,1986) we just did a qm on the trans for the charges and assumed that the other parameters would lead to a reasonable cis/trans energy difference.
For Cornell et al. (1995), we did a 4 amino acid fit for proline which included 2 conformations each for both trans and cis and cis isomers. Then within each of those sets there were two different ring puckers. This is described on page 1361 of the JCC article (Cieplak, Cornell, Bayly, and Kollman, vol 16, pp 1357-1377 (1995)).
The new force field does not use hydrogen bonding terms.