{
  "context": "Tests whether the agent reviews TypeScript changes with the skill's priorities: behavioral risks, contracts, validation, async and state bugs, tests, and type soundness. The task asks for a review artifact rather than implementation.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Findings first",
      "description": "The review starts with concrete findings rather than a summary-first response.",
      "max_score": 8
    },
    {
      "name": "Behavior focus",
      "description": "Findings prioritize behavioral risks, broken contracts, weak validation, async or state bugs, missing tests, or type unsoundness over style-only comments.",
      "max_score": 12
    },
    {
      "name": "Line grounded",
      "description": "Each finding cites a specific file and line or tight line range from the provided code.",
      "max_score": 8
    },
    {
      "name": "Boundary validation",
      "description": "Calls out unchecked runtime boundary data where static TypeScript types do not protect request, config, or external response values.",
      "max_score": 10
    },
    {
      "name": "Contract risk",
      "description": "Calls out any public contract, persisted shape, event payload, or exported API change that is not explicit or tested.",
      "max_score": 8
    },
    {
      "name": "Async risk",
      "description": "Calls out floating promises, swallowed errors, hidden races, missing timeout or cancellation, or misuse of Promise.all when present.",
      "max_score": 8
    },
    {
      "name": "Type soundness",
      "description": "Calls out any, unsafe assertions, broad primitives, vague null or undefined failures, or duplicated schema/type contracts when present.",
      "max_score": 8
    },
    {
      "name": "Security risk",
      "description": "Calls out relevant authorization, tenant filtering, unsafe SQL, unsafe HTML, redirect/path/URL, or secret exposure risks when present.",
      "max_score": 8
    },
    {
      "name": "Test gap",
      "description": "Identifies missing focused tests for parsing, permissions, migrations, state transitions, bug fixes, or public behavior as applicable.",
      "max_score": 8
    },
    {
      "name": "Severity order",
      "description": "Orders findings by severity, with the highest-risk correctness or security issue before lower-risk maintainability issues.",
      "max_score": 6
    },
    {
      "name": "No style nitpick",
      "description": "Does not include style-only observations unless they are tied to a concrete maintenance or behavioral risk.",
      "max_score": 6
    },
    {
      "name": "Open questions",
      "description": "Separates open questions or assumptions from findings and keeps any change summary secondary.",
      "max_score": 10
    }
  ]
}