vitest-evals

vitest-evals is a Vitest extension that enables developers to run LLM evaluations (evals) directly within their existing test suite, treating evals as first-class test cases. It’s designed for AI engineers and full-stack developers who want to validate model outputs, prompt quality, or agent behavior using familiar Vitest assertions and workflows. Its interesting because it bridges the gap between traditional software testing and AI evaluation, allowing teams to catch regressions and iterate on prompts with the same rigor as unit tests.