Refactor-Monolith-Scheduler

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 refactor, rationale, tests. Refactor a scheduler monolith into parsing, planning, and execution modules. Preserve retry and cancellation semantics.

{
  "metrics": {
    "visible": {
      "checks": [
        {
          "contains": [
            "parsing"
          ]
        },
        {
          "contains": [
            "planning"
          ]
        },
        {
          "contains": [
            "execution"
          ]
        }
      ]
    },
    "hidden": {
      "checks": [
        {
          "contains": [
            "retry"
          ]
        },
        {
          "contains": [
            "cancellation"
          ]
        },
        {
          "contains": [
            "scheduler"
          ]
        }
      ]
    },
    "intent": {
      "checks": [
        {
          "contains": [
            "decomposition"
          ]
        },
        {
          "contains": [
            "modernization"
          ]
        },
        {
          "contains": [
            "monolith"
          ]
        }
      ]
    }
  }
}