Generally, upon any radical vagrant-spk changes, I test against php-app-to-package-for-sandstorm and python-app-to-package-for-sandstorm, which are both excellent baseline test apps that use the base stack scripts for lemp and uwsgi respectively. (lesp is basically a simpler version of the lemp stack, so I also feel it is adequately tested this way.)
However, I do not have good test cases for our other stacks, which occasionally means they are broken until someone has a bad experience trying to use them. I think we need "Hello, world" level test apps that run with unmodified (or minimally modified) stack scripts for the Node, Meteor, and Go stacks.