One embodiment of the present invention sets forth a method for generating one or more tests for testing a command provided by a command line interface. The command is specified in a pre-determined meta-language. The meta-language is then converted in the tokenized post-fix notation, which is processed to generate a direct acyclic graph (DAG) that represents the different permutations of the command. By traversing the DAG from a root node to a leaf node, a test can be generated. The test is then executed within the command line interface, and the result analyzed to determine whether the command line interface is operating correctly.