Specifically creating whole commands (triggers, actions, and configuration): practically speaking, no.
If you're just looking to have phrases recognized by the speech recognition system, that could be done using a token and the "Wait For Spoken Response" action.
You could pass the information from your Python script using the "-PassedText" command line parameter, for example.
There is also an undocumented feature that allows the use of token in command phrases, which you could utilize, but that requires reloading the profile to apply any changes.