Sec-Secret-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 a secret scanner that only matches API_KEY= and misses private keys, bearer tokens, and high-entropy assignments.

{
  "metrics": {
    "correct": {
      "checks": [
        {
          "contains": [
            "secret"
          ]
        },
        {
          "contains": [
            "private key"
          ]
        },
        {
          "contains": [
            "bearer"
          ]
        },
        {
          "contains": [
            "entropy"
          ]
        },
        {
          "contains": [
            "token"
          ]
        }
      ]
    },
    "hidden": {
      "checks": [
        {
          "contains": [
            "regex"
          ]
        },
        {
          "contains": [
            "false positive"
          ]
        },
        {
          "contains": [
            "redact"
          ]
        },
        {
          "contains": [
            "rotation"
          ]
        },
        {
          "contains": [
            "allowlist"
          ]
        }
      ]
    }
  }
}