{
  "scenario": "smoke-metadata-fail",
  "sidecar": "callback-recorded-PASS.json",
  "judge_model": "openai/gpt-4o-mini",
  "duration_seconds": 31.35,
  "idle_disconnected": false,
  "rubric_results": [
    {
      "name": "end_call_marker",
      "score": 1.0,
      "success": true,
      "threshold": 0.5,
      "reason": "The conversation includes the tool-call marker '[end_call: callback_recorded]' in the assistant's response, indicating that the criteria for tool-call presence are fully met. The assistant successfully recorded the user's contact information and confirmed the callback, demonstrating effective communication and adherence to the evaluation steps."
    }
  ],
  "metadata_results": [
    {
      "field": "duration_seconds",
      "op": "gt",
      "expected": 100,
      "actual": 31.35,
      "success": false,
      "reason": "expected gt 100, got 31.35"
    },
    {
      "field": "frames_emitted",
      "op": "between",
      "expected": [
        50,
        500
      ],
      "actual": 264,
      "success": true,
      "reason": "expected between 50..500 (inclusive), got 264"
    }
  ],
  "passed": false
}