From the first check_nrpe line in the link:
check_nrpe -H 192.168.142.137 -t 30 -c CheckMEM...
That means its going to connect to the remote host (.137) on port 5666 by default and run the CheckMEM command defined in the remote hosts's nrpe.cfg. The command defines the script or binary executed on the remote host to get the data. The script or binary has to be on the remote host. This (transferring the script/binary to a bunch of remote hosts) can be a pain after a while, as can keeping that nrpe.cfg current.
When possible, pull data from the hosts rather than use NRPE--its less trouble to update. There are articles written on using puppet, etc. for keeping things up to date, so if you have a bunch of hosts look for that.
Look for Monitoring_Hosts_Using_NRPE.pdf
Regarding your example, I'd say no. NRPE is for running commands remotely.
commands.cfg is modified by adding/modifying commands in CCM.
There are lots of plugins defined by default that do most of what you want, but you may want to get some that aren't installed, or create some yourself. Google nagios plugins.
For the memory check you describe, try this:
# locate nrpe.cfg
/usr/local/nagios/etc/nrpe.cfg
# grep custom_check_mem /usr/local/nagios/etc/nrpe.cfg
# grep custom_check_mem /usr/local/nagios/etc/nrpe/common.cfg
command[check_mem]=/usr/local/nagios/libexec/custom_check_mem -n $ARG1$
common.cfg contains more commands, and is sourced by nrpe.cfg and check_mem is the command, while /usr/local/nagios/libexec/custom_check_mem is a script.
# file /usr/local/nagios/libexec/custom_check_mem
/usr/local/nagios/libexec/custom_check_mem: Bourne-Again shell script text executable
Your nagios service would use check_nrpe locally to ask the remote NRPE to run check_mem on the remote host, which then executes custom_check_mem and returns the exit status, and text output.
You can use CCM to define a service something like this:
check_nrpe -H <host> -t 30 -c check_mem <args>...
Just go to CCM, services, click add new, and select check_nrpe. Look at the args it wants. $ARG1$ will be check_mem, $ARG2$ other needed args.
Give it a name and a host to run on, and test.
I'd suggest using templates and hostgroups as you go, since trying to switch to a maintainable system after building one the other way is a hassle. I had to copy all my config files, create a few scripts to remove junk, delete things, then import them to get things right. Why?
Last night I needed to change the host check to work w/ IPv6, since the existing check didn't work w/ hosts added last night. I was able to change it in one template and update the host check on ~200 hosts...