Reason-State-Cart-Discount-Order

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 cart applies a 10% coupon, then a $5 store credit, then tax. Another implementation applies credit before percentage discount. Determine which order matches the stated policy and why totals differ.

{
  "metrics": {
    "accuracy": {
      "checks": [
        {
          "contains": [
            "coupon"
          ]
        },
        {
          "contains": [
            "store credit"
          ]
        },
        {
          "contains": [
            "tax"
          ]
        }
      ]
    },
    "evidence": {
      "checks": [
        {
          "contains": [
            "order"
          ]
        },
        {
          "contains": [
            "totals differ"
          ]
        },
        {
          "contains": [
            "policy"
          ]
        }
      ]
    },
    "constraint": {
      "checks": [
        {
          "contains": [
            "10%"
          ]
        },
        {
          "contains": [
            "before"
          ]
        },
        {
          "contains": [
            "credit"
          ]
        }
      ]
    }
  }
}