Reasoning-Uad-False-Positive-Alert

The model receives the prompt (and optional system message). The run uses scorer rubric_json_metrics with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.

Return JSON only with keys answer, evidence, constraints. An alert fires after one failed login from a new country, but the user is traveling and MFA succeeded. Explain the uncertainty and safest next action.

{
  "metrics": {
    "accuracy": {
      "checks": [
        {
          "contains": [
            "traveling"
          ]
        },
        {
          "contains": [
            "MFA"
          ]
        },
        {
          "contains": [
            "uncertainty"
          ]
        }
      ]
    },
    "evidence": {
      "checks": [
        {
          "contains": [
            "failed login"
          ]
        },
        {
          "contains": [
            "new country"
          ]
        },
        {
          "contains": [
            "alert"
          ]
        }
      ]
    },
    "constraint": {
      "checks": [
        {
          "contains": [
            "false positive"
          ]
        },
        {
          "contains": [
            "investigate"
          ]
        },
        {
          "contains": [
            "do not lock solely"
          ]
        }
      ]
    }
  }
}