The driving force behind it is that monitoring shouldn't fit into a mold. My thinking is that you should write a script to perform whatever mission critical thing your application provides, and monitor the output of that.

Thanks for the feedback!