Score an OpenAPI 3.x or Swagger 2.x spec on agent-readiness — i.e. does an LLM-driven caller have what it needs to call the API correctly without guessing. Returns a 0..100 score, severity counts, and a structured list of violations with stable ru...