Sec-Xss-Reflected-Output

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. A web handler concatenates user input directly into an HTML response string without any escaping or encoding before sending it to the browser.

{
  "metrics": {
    "correct": {
      "checks": [
        {
          "contains": [
            "reflected XSS"
          ]
        },
        {
          "contains": [
            "user input"
          ]
        },
        {
          "contains": [
            "HTML escaping"
          ]
        },
        {
          "contains": [
            "sanitize"
          ]
        },
        {
          "contains": [
            "encode"
          ]
        }
      ]
    },
    "hidden": {
      "checks": [
        {
          "contains": [
            "Content-Security-Policy"
          ]
        },
        {
          "contains": [
            "output encoding"
          ]
        },
        {
          "contains": [
            "context-aware"
          ]
        },
        {
          "contains": [
            "trusted types"
          ]
        },
        {
          "contains": [
            "template engine"
          ]
        }
      ]
    }
  }
}