I'm doing something similar but not for an emulator, using tests other projects in a similar space use to validate things.
I also know a friend that feeds Claude Gherkin formatted test criteria, and apparently that yields insanely good Claude output.
Claude Gherkin
Any good recommended reading?https://cucumber.io/docs/gherkin/
Very neat, seems quite appropriate for the way people currently work with LLMs.
I guess humans need their oracles too: teachers, encyclopaedias, degrees, phds.
Bravo, awesome project, really creative!