websims-sd.pages.dev

studio_run_evaluation

Run the bounded correctness search and/or the replicated performance measurement for one version. Refused while the project's workload is still the Studio placeholder, because the first result locks the yardstick for every version. Performance defaults off and is refused for a repository model until every component and link has observed runtime or user performance evidence. Verdicts: VIOLATED, NO_VIOLATION_WITHIN_BOUNDS, INCONCLUSIVE_BOUND_REACHED, INVALID_MODEL. NO_VIOLATION_WITHIN_BOUNDS is not proof of safety, and INCONCLUSIVE_BOUND_REACHED establishes nothing either way. Fetch the counterexample with studio_get_evaluation.

Input schema

{
  "type": "object",
  "properties": {
    "candidateId": {
      "type": "string",
      "minLength": 1,
      "maxLength": 64
    },
    "correctness": {
      "type": "boolean",
      "default": true
    },
    "performance": {
      "type": "boolean",
      "default": false
    }
  },
  "additionalProperties": false,
  "required": [
    "candidateId"
  ]
}