Making JavaScript-rendered pages crawlable with build-time pre-rendering and auto-noindex for thin pages

A Korean information site generated as static HTML had sections drawn by JavaScript from a JSON file, leaving crawlers with almost no readable text. The author pre-renders that same JSON into the HTML at build time, so crawlers and users see equivalent content. Two further fixes followed: a lastmod date that changes only when page content actually changes, and a pass that marks thin generated pages noindex while keeping their links followable. The write-up includes the relevant build-script code and notes known weaknesses, such as the date-hashing approach ignoring fraction changes. Some interactive sections remain thin because their output depends on user input.