A free Claude code that scores your AI agents output against the rubric that top-tier consulting companies use to evaluate their workflows. Catches the two things that generic evals miss: sycophancy, and silent ambiguity, No API keys, no SDKs, no signup. One command to install MIT licensed.
