Most people don't know that they ask basic prompts when interacting with LLMs. But there is a BETTER way.
Simply add role and 10x the output of the responses you get.
I don't know if the test I wrote works? Maybe I should write a test that tests the test is working. And while I am at it, why not create a test that test if the test that tests the test, the original test, works? I should test that.