{
  "context": "Evaluates how well the solution uses Python's standard library concurrency APIs (asyncio + subprocess) to run shell commands in parallel with limits, timeouts, and fail-fast cancellation while keeping ordered results. Checks focus on asyncio subprocess usage, concurrency gating, timeout enforcement, cancellation, and ordered aggregation.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Async subprocess",
      "description": "Commands are spawned with asyncio-aware subprocess APIs (e.g., asyncio.create_subprocess_exec or asyncio.create_subprocess_shell) and stdout/stderr are captured via PIPE; avoids blocking subprocess.run/call.",
      "max_score": 30
    },
    {
      "name": "Concurrency limit",
      "description": "No more than max_concurrency commands run at once by using asyncio.Semaphore/TaskGroup gating with asyncio.create_task/gather instead of firing all tasks unchecked.",
      "max_score": 25
    },
    {
      "name": "Timeout handling",
      "description": "Per-command timeouts use asyncio.wait_for (or equivalent) around process completion; on asyncio.TimeoutError the process is terminated (kill/terminate) and awaited to avoid zombies, with timed_out flagged and return_code set to None.",
      "max_score": 20
    },
    {
      "name": "Fail-fast cancellation",
      "description": "When fail_fast is enabled and a command returns non-zero, remaining tasks are cancelled via Task.cancel/asyncio.wait and CancelledError handling; their subprocesses are cleaned up and results reported with return_code None and timed_out False.",
      "max_score": 15
    },
    {
      "name": "Ordered results",
      "description": "Results preserve the original command order by indexing tasks (e.g., storing input index with each asyncio.Task and reassembling after gather/as_completed).",
      "max_score": 10
    }
  ]
}