Sec-Auth-Log-Anomaly-Detector

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 verdict, risk, mitigation. Review login anomaly detection that counts failures globally but not per account, IP, or time window.

{
  "metrics": {
    "correct": {
      "checks": [
        {
          "contains": [
            "rate"
          ]
        },
        {
          "contains": [
            "account"
          ]
        },
        {
          "contains": [
            "IP"
          ]
        },
        {
          "contains": [
            "time window"
          ]
        },
        {
          "contains": [
            "anomaly"
          ]
        }
      ]
    },
    "hidden": {
      "checks": [
        {
          "contains": [
            "credential stuffing"
          ]
        },
        {
          "contains": [
            "threshold"
          ]
        },
        {
          "contains": [
            "sliding window"
          ]
        },
        {
          "contains": [
            "alert"
          ]
        },
        {
          "contains": [
            "lockout"
          ]
        }
      ]
    }
  }
}