Reason-State-Lru-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. An LRU cache of capacity 3 receives: put A, put B, put C, get A, put D. Determine the keys from least to most recent.

{
  "metrics": {
    "accuracy": {
      "checks": [
        {
          "contains": [
            "B"
          ]
        },
        {
          "contains": [
            "C"
          ]
        },
        {
          "contains": [
            "A"
          ]
        },
        {
          "contains": [
            "D"
          ]
        }
      ]
    },
    "evidence": {
      "checks": [
        {
          "contains": [
            "get A"
          ]
        },
        {
          "contains": [
            "capacity 3"
          ]
        },
        {
          "contains": [
            "least"
          ]
        }
      ]
    },
    "constraint": {
      "checks": [
        {
          "contains": [
            "stateful"
          ]
        },
        {
          "contains": [
            "order"
          ]
        },
        {
          "contains": [
            "evict B"
          ]
        }
      ]
    }
  }
}