portyu9 / qa-automation-ai-agent-evals

git clone https://github.com/portyu9/qa-automation-ai-agent-evals.git

A provider-neutral quality-engineering framework for evaluating autonomous agents by observable outcomes, side effects, authority boundaries, approval intent, adversarial conditions, verified evaluation preconditions, protocol state, authorization behavior, reliability, and reproducible evidence—not by persuasive final prose.