Extract Condition

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 allow_login(user):
    if user["active"] and user["verified"] and not user["locked"]:
        return True
    return False

{
  "expected_contains": [
    "return user[\"active\"] and user[\"verified\"] and not user[\"locked\"]",
    "is_allowed"
  ]
}