When you have a set of "standard" things that you monitor on a heterogenous set of servers, and suddenly busybox-based machines are added to that set,
you will likely find that you don't have a compiler handy and that a large number of Unix commands are unavailable as well.
In other words, you won't have working standard plugins available.
Rather than going on a quest for a (cross?-)compiler, I reimplemented the most important plugins as shell scripts, in spite of the limited set of commands available.
check_disk, check_load and check_swap will plain ignore all but the most usual options.
(So you can run them with your standard $ARG$s without getting an error returned.)
check_procs does that, too, but it also has the new options -B and --busybox-sh, which extend the functionality of -C.
On a run-of-the-mill Linux system, the executable of a "syslogd" will in fact be named "syslogd"; on a busybox system, "syslogd" actually is a symlink to "busybox".
Checking with "-B syslogd" instead of "-C syslogd" will give you a match nonetheless.
Likewise, --busybox-sh will find running Bourne shell scripts for you, as /bin/sh is a symlink to "busybox" as well.