Please submit all questions to the LAMMPS mailing list. You proposed solution does not make any sense, because you are only running 1 step of MD. What you need to do isrun an NVT or NVE simulation to generate an equilibrated distribution of configurations and estimate the average stress tensor from that distribution. Then repeat the process after apply a strain. The change in each stress component gives you an estimate of the corresponding elastic constant. Because of the many factors involved in generating good statistics, there is no easy way to automate this process in a black box script; it must be done by hand.
Aidan