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 safe_name(user):\n    name = user.get('name')\n    if name is None:\n        return ''\n    name = name.strip()\n    if not name:\n        return ''\n    return name\n

{
  "expected_contains": [
    "if name is None",
    "name.strip()",
    "return \"\""
  ]
}