Dict Get

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.

def city(user):
    if "address" in user and "city" in user["address"]:
        return user["address"]["city"]
    return ""

{
  "expected_contains": [
    "get(\"address\"",
    ".get(\"city\""
  ]
}