Cleanup

The model receives the prompt (and optional system message). The run uses scorer contains_any with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.

Refactor without changing behavior. Return only Python code.\n\ndef order_total(order):\n    total = 0\n    for line in order['lines']:\n        total += line['qty'] * line['price']\n    return total\n

{
  "expected_contains": [
    "for line in order[\"lines\"]",
    "line[\"qty\"] * line[\"price\"]"
  ]
}