I could.

Applying an "automask" for an EPI dataset:
3dAutomask -prefix amask epi_r1+orig
3dcalc -a epi_r1+orig -b amask+orig. -expr 'a*b' -prefix epi_r1_am

Using a skull-stripped anatomical dataset as a mask for an EPI dataset (assumes anatomical dataset and EPI are already in good alignment):
3dSkullStrip -input ./anat+orig -prefix anat_ns
3dresample -master epi_r1+orig -prefix anat_ns_rs -inset anat_ns+orig
3dcalc -a epi_r1+orig -b anat_ns_rs+orig -expr 'a*step(b)' -prefix epi_r1_ns

Using align_epi_anat.py to skullstrip and align anatomical dataset first :
align_epi_anat.py -anat anat+orig -epi epi_r1+orig -epi_base 0 -suffix _alepi
3dresample -master epi_r1+orig -prefix anat_alepi_rs -input anat_alepi+orig
3dcalc -a epi_r1+orig -b anat_alepi_rs+orig. -expr 'a*step(b)' -prefix epi_r1_al_ns

3dSkullStrip can also be used to mask the EPI data directly. The normal output for 3dSkullStrip is a volume slightly normalized from the original volume. You can use options like -orig_vol to get the original intensity values or -mask_vol to get a mask volume to apply to the original data. Also consider using afni_proc.py for creating a processing pipeline for fMRI. It includes various options for alignment and automasking of the data.