"pytest for AI agents." Open-source testing framework. Define what your agent should do. Run it. Get pass/fail results with AI diagnosis. Tests whether your agent behaves correctly across runs
