Fix Sum Start

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

Fix the bug and return corrected Python code only.

def sum_numbers(nums):
    total = 1
    for n in nums:
        total += n
    return total

{
  "expected_contains": [
    "total = 0",
    "for n in nums",
    "return total"
  ]
}