End-to-End Evaluation does not support policies that don't generate utters

I am currently in the planning phase of E2E testing Rasa core.

However, it does not seem that Rasa E2E testing supports cases where something else than utters are generated. Is this true, and if so, are there work arounds?

Welcome to the forum. End to end testing will predict the actions however it will not actually invoke the action.

Running rasa test --stories test/test_stories.md --e2e on the following will work:

## corny_joke
* joke: tell me a joke
  - utter_ask_joke_type
* joke: give me a [corny](joke_type) joke
  - slot{"joke_type":"corny"}
  - action_corny