{
  "context": "Tests whether the agent correctly handles round-trip verification failures as critical tier, immediately halts further autonomous processing, escalates without retry, and never suppresses data-loss signals.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Round-trip verification",
      "description": "Code performs a read-back check after a write and compares the result to the expected written value",
      "max_score": 10
    },
    {
      "name": "Critical tier classification",
      "description": "Code or comments label the read-back mismatch as 'critical' (or highest severity equivalent) — not operational or cosmetic",
      "max_score": 12
    },
    {
      "name": "Halt autonomous processing",
      "description": "On verification failure, code stops further processing for this path (raises exception, returns error state, or otherwise prevents continuation) rather than logging and continuing",
      "max_score": 12
    },
    {
      "name": "Immediate escalation",
      "description": "On verification failure, code escalates immediately (calls escalation function, raises, or sets escalation flag) without retry attempts first",
      "max_score": 10
    },
    {
      "name": "No retry on data-loss",
      "description": "Code does NOT retry the write operation after a verification mismatch is detected (no retry loop or fallback write on mismatch)",
      "max_score": 12
    },
    {
      "name": "No suppression of data-loss",
      "description": "Code does NOT swallow or suppress the verification failure (no bare except, no silent continue, no suppression budget applied to this failure type)",
      "max_score": 10
    },
    {
      "name": "Evidence in report",
      "description": "Report structure includes the failure signal AND observed evidence (e.g., expected vs actual values, or equivalent mismatch information)",
      "max_score": 8
    },
    {
      "name": "Escalation status in report",
      "description": "Report structure includes escalation status (e.g., a field indicating escalated=True or equivalent)",
      "max_score": 8
    },
    {
      "name": "Tests present",
      "description": "test_write_guard.py contains at least 2 test cases covering the verification failure scenario",
      "max_score": 10
    },
    {
      "name": "Tier label in code",
      "description": "Code includes an explicit string or constant referencing the tier (e.g., 'critical' appears as a value or comment in the classification logic)",
      "max_score": 8
    }
  ]
}