Reason-Constraint-Maintenance-Window

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. A deployment needs 45 minutes, the maintenance window is 01:00-01:30 UTC, and rollback requires 10 minutes. Decide whether the deployment can proceed under the constraints.

{
  "metrics": {
    "accuracy": {
      "checks": [
        {
          "contains": [
            "cannot proceed"
          ]
        },
        {
          "contains": [
            "45 minutes"
          ]
        },
        {
          "contains": [
            "30"
          ]
        }
      ]
    },
    "evidence": {
      "checks": [
        {
          "contains": [
            "rollback"
          ]
        },
        {
          "contains": [
            "maintenance window"
          ]
        },
        {
          "contains": [
            "UTC"
          ]
        }
      ]
    },
    "constraint": {
      "checks": [
        {
          "contains": [
            "rollback requires 10 minutes"
          ]
        },
        {
          "contains": [
            "window"
          ]
        },
        {
          "contains": [
            "breach"
          ]
        }
      ]
    }
  }
}