If we were to begin running these regularly against puppet and recording the results, we could use the results of those runs to spot regressions before they ship and before they impact other components further up the stack such as puppetserver where it can be more costly to diagnose.