AI citation readiness grader

The rubric is a fixed weighted read, not a model score: sourced claims (0.20), visible dates (0.15), named authors (0.15), self-contained answer blocks (0.15), stable entity facts (0.10), question-shaped headings (0.10), a direct answer near the top (0.10), and a clean heading ladder (0.05). Each criterion quotes the text it judged, capped at 4 quotes, and grading covers at most 400 paragraphs. This measures how citable the pasted text is; it cannot promise that an answer engine will cite the page, and it never fetches a URL, so content that only exists after scripts run needs the rendered HTML pasted here.