How do you approach mocking?

[afg1]

I'm trying to write some tests for agent-ish bits of guidance code - specifically some flow control that is executed depending on the output of the LLM doing a select:

    with user():
        lm += (f"Here is a statement about the gene {gene_symbol}: \n{answer}\n"
               "Does this statement make an association between the gene and a disease? "
               "Answer only Yes or No\n")
    with assistant():
        lm += select(["Yes", "No"], name="disease_mention") + '\n'
    if lm["disease_mention"] == "Yes":
        ... things happen

When I use the mock LLM in https://github.com/guidance-ai/guidance/blob/855ce5bba90ad73d2f687e67a5abb1d4f00bee29/guidance/models/_mock.py I don't think I ever get the LLM to say 'Yes'. How do others approach this? I'd managed to mock/monkeypatch bits of the mock LLM to test other things like context exhaustion, but I can't see an obvious way to force something like this. Any help would be very much appreciated!