Use Comprehension

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 positive_even_squares(nums):
    out = []
    for n in nums:
        if n > 0:
            if n % 2 == 0:
                out.append(n * n)
    return out

{
  "expected_contains": [
    "[n * n for n in nums",
    "if n > 0 and n % 2 == 0",
    "return ["
  ]
}