{
  "context": "Tests whether the agent produces an architecture audit report with the correct section structure, proper finding block format, correct improvement sequence shape, and avoids forbidden report elements like scorecards and strengths sections.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Section: Repo Brief",
      "description": "Report contains a '## Repo Brief' section",
      "max_score": 4
    },
    {
      "name": "Section: Highest-Leverage Issues",
      "description": "Report contains a '## Highest-Leverage Structural Issues' section",
      "max_score": 4
    },
    {
      "name": "Section: Boundary and Layering",
      "description": "Report contains a '## Boundary and Layering Problems' section",
      "max_score": 4
    },
    {
      "name": "Section: Dependency and Composition",
      "description": "Report contains a '## Dependency and Composition Problems' section",
      "max_score": 4
    },
    {
      "name": "Section: Testability and Change Friction",
      "description": "Report contains a '## Testability and Change Friction' section",
      "max_score": 4
    },
    {
      "name": "Section: Improvement Sequence",
      "description": "Report contains a '## Improvement Sequence' section",
      "max_score": 4
    },
    {
      "name": "Section: Open Evidence Gaps",
      "description": "Report contains a '## Open Evidence Gaps' section",
      "max_score": 4
    },
    {
      "name": "Section order",
      "description": "All 7 required sections appear in the exact order: Repo Brief, Highest-Leverage Structural Issues, Boundary and Layering Problems, Dependency and Composition Problems, Testability and Change Friction, Improvement Sequence, Open Evidence Gaps",
      "max_score": 8
    },
    {
      "name": "Highest-Leverage as summary",
      "description": "The '## Highest-Leverage Structural Issues' section is a short prioritized summary (not full finding blocks with Severity/Evidence/Why fields) — it summarizes top root issues rather than listing detailed findings",
      "max_score": 6
    },
    {
      "name": "Finding block: Severity field",
      "description": "At least one finding block in the category sections includes a 'Severity:' field with value critical, high, or medium",
      "max_score": 8
    },
    {
      "name": "Finding block: Evidence field",
      "description": "At least one finding block includes an 'Evidence:' field citing specific files, modules, or code surfaces",
      "max_score": 8
    },
    {
      "name": "Finding block: Why it matters",
      "description": "At least one finding block includes a 'Why it matters:' field",
      "max_score": 8
    },
    {
      "name": "Finding block: Recommended improvement",
      "description": "At least one finding block includes a 'Recommended improvement:' field",
      "max_score": 8
    },
    {
      "name": "Finding block: Likely refactor surface",
      "description": "At least one finding block includes a 'Likely refactor surface:' field",
      "max_score": 6
    },
    {
      "name": "Improvement Sequence format",
      "description": "The '## Improvement Sequence' section uses numbered steps each with a 'Goal:' line and a 'Why first:' line",
      "max_score": 8
    },
    {
      "name": "No scorecards",
      "description": "Report does NOT include a scorecard, numerical ratings table, or score summary (e.g. no 'Score: X/10' or rating grid)",
      "max_score": 6
    },
    {
      "name": "No strengths section",
      "description": "Report does NOT contain a 'Strengths', 'What's Working Well', or 'Positives' section heading",
      "max_score": 6
    }
  ]
}