I think it’d be useful to have the option to have the build fail (possibly via a separate Github status?) if the number of tests run decreases compared to the last comparable hash. This can be useful especially when debugging parallelization logic as a failsafe to ensure you didn’t forget to delegate e.g. an rspec file to one of the test runners.
Otherwise, it could be useful a hygiene approach for a project, encouraging people to generally increase the test coverage as time goes on without doing a strict code coverage comparison.