This page needs work but here's the gist of what you need to know.
The test program is in tests/nrTEST
If you have cloned the repository and you cd into tests, then you can run:
NUMEROUSAPIKEY=~/.numerous ./nrTEST -D
The "-D" flag will make the test program execute using the local numerous.py library (in .. from the tests directory) and the local nr.py shell command (in ../shell-cmd from the tests directory) and it will run the tests twice - first with python version 2 and then with python version 3.
You can use the "-Q" flag to run "quick". It's not really quick but it does skip a few of the more time-intensive tests.
You can use the "-C" flag to "continue" even after certain errors. The atomicity of ADD at the Numerous server still isn't 100% (as of this writing) and so -C allows that part of the test suite to fail without aborting the rest of the test.
You can use the "-4" flag to skip the throttling tests which are especially time consuming. This is implied by -Q (so -4 skips a subset of what -Q skips).
My normal mode for testing any changes is: ./nrTEST -CDQ
and every now and then I'll leave out the -Q flag just to make sure I
haven't broken anything that is tested that way. If you leave out the
-C then your tests may fail if the numerous people haven't fixed the
current atomicity bugs in ADD.
It might be best to have a Numerous API Key for a "test" account because this will end up making a lot of metrics in the account during the test (though it does try to clean them up; frankly I just use my normal account and every now and then I just have to clean it up by hand if the test went awry).