{
  "context": "The agent audits a Next.js landing page from source code only (no live URL available). The audit should follow a structured workflow: context gathering, five specialist lenses, synthesis, and a complete report. Criteria verify whether the agent follows the skill's specific instructions for report format, finding structure, specialist coverage, and analysis guardrails.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Copy & Messaging section",
      "description": "The report contains a section analyzing copy and messaging (headlines, CTA wording, value proposition, messaging framework, or reading level)",
      "max_score": 4
    },
    {
      "name": "CRO & UX section",
      "description": "The report contains a section analyzing conversion and UX elements (form design, CTA placement, navigation, social proof placement, or distraction paths)",
      "max_score": 4
    },
    {
      "name": "Brand & Positioning section",
      "description": "The report contains a section analyzing brand clarity, differentiation, trust signals, or positioning",
      "max_score": 4
    },
    {
      "name": "Technical Performance section",
      "description": "The report contains a section analyzing technical signals (page speed indicators, image attributes, analytics scripts, SEO metadata, or Core Web Vitals readiness)",
      "max_score": 4
    },
    {
      "name": "Funnel Alignment section",
      "description": "The report contains a section analyzing funnel stage fit, audience temperature, page type classification, or post-conversion experience",
      "max_score": 4
    },
    {
      "name": "Finding: label present",
      "description": "At least three individual findings use the exact label 'Finding:' (case-insensitive) at the start of the finding block",
      "max_score": 8
    },
    {
      "name": "Evidence: label present",
      "description": "At least three individual findings use the exact label 'Evidence:' (case-insensitive), citing a specific code element, file reference, or quoted text",
      "max_score": 8
    },
    {
      "name": "Why it matters: label present",
      "description": "At least three individual findings use the exact label 'Why it matters:' (case-insensitive)",
      "max_score": 7
    },
    {
      "name": "Suggested fix: label present",
      "description": "At least three individual findings use the exact label 'Suggested fix:' (case-insensitive)",
      "max_score": 7
    },
    {
      "name": "Severity: label present",
      "description": "At least three individual findings use the exact label 'Severity:' (case-insensitive) with a value of critical, high, medium, or low",
      "max_score": 5
    },
    {
      "name": "What's Working Well section",
      "description": "The report includes a section explicitly titled 'What's Working Well' (or equivalent such as 'Strengths') that cites specific evidence from the source files — not a generic placeholder",
      "max_score": 12
    },
    {
      "name": "Severity-sorted issue list",
      "description": "The report contains a prioritized issue table or list where Critical and High severity items appear before Medium and Low items",
      "max_score": 10
    },
    {
      "name": "Narrative executive summary",
      "description": "The executive summary is written in paragraph form (at least two sentences of prose) and does NOT consist solely of bullet points",
      "max_score": 10
    },
    {
      "name": "Evidence gaps stated",
      "description": "The report explicitly acknowledges that the live page was not accessible and analysis is based on source code, or equivalently states a limitation about what evidence was unavailable",
      "max_score": 5
    },
    {
      "name": "Cold-traffic assumption",
      "description": "The report explicitly states a cold-traffic assumption or notes that traffic source is unknown (since no UTM, campaign copy, or analytics tracking was found in the source code)",
      "max_score": 5
    },
    {
      "name": "No visual hallucinations",
      "description": "The report does NOT make claims about visual aesthetic quality (e.g., 'the design looks clean', 'colors are well-chosen', 'whitespace feels adequate') that are not backed by specific CSS values, class names, or DOM structure evidence",
      "max_score": 3
    }
  ]
}