Reason-State-Token-Refresh

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 client refreshes an access token at T+55 minutes, but two tabs do it at the same time and one overwrites storage with an older token response. Explain the state race.

{
  "metrics": {
    "accuracy": {
      "checks": [
        {
          "contains": [
            "two tabs"
          ]
        },
        {
          "contains": [
            "older token"
          ]
        },
        {
          "contains": [
            "overwrite"
          ]
        }
      ]
    },
    "evidence": {
      "checks": [
        {
          "contains": [
            "storage"
          ]
        },
        {
          "contains": [
            "race"
          ]
        },
        {
          "contains": [
            "refresh"
          ]
        }
      ]
    },
    "constraint": {
      "checks": [
        {
          "contains": [
            "compare issued time"
          ]
        },
        {
          "contains": [
            "single flight"
          ]
        },
        {
          "contains": [
            "stale"
          ]
        }
      ]
    }
  }
}