Stub LLM Calls with Gem RubyLLM::Test

[toddkummer]

If you have application logic that interacts with an LLM, it can be hard to test. After having iterated through a few versions, I just released the RubyLLM::Test gem that allows you to stub cleanly and easily.

RubyLLM::Test.stub_response("Outlook good")

chat = RubyLLM.chat
response = chat.ask "What are the odds this works?"

assert_equal "Outlook good", response.content

You can stub multiple calls, stub in blocks, pass in a RubyLLM::Message for full control, or pass in a hash to get a JSON response.

https://github.com/RockSolt/ruby_llm-test