When running nirt from the mged command line it is not possible to alter the behavior of the pre-canned settings to nirt because the user command line arguments are added after the "s" command to shoot is added to the nirt command string (see "src/libged/nirt.c").
This can be reproduced by running nirt with a modified output format:
mged> set glob_compat_mode 0
mged> nirt -e {fmt r "start (%g %g %g)\n" x_orig y_orig z_orig}
Arguably, the user command line arguments should be appended to the mged defaults prior to the first "s" shoot command (see src/libged/nirt.c).
At the moment, it is possible to modify the nirt state and shoot a second ray to get the output desired from the second ray (note additional "s" command appended):
mged> set glob_compat_mode 0
mged> nirt -e {fmt r "start (%g %g %g)\n" x_orig y_orig z_orig ; s}
The output from the first nirt ray must be ignored, which is rather awkward.
Hey! I just tried to solve it, though it is not working right now.