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 count_enabled(flags):\n    c = 0\n    for f in flags:\n        if f == True:\n            c += 1\n    return c\n

{
  "expected_contains": [
    "sum(",
    "for f in flags"
  ]
}