docs-from-sitelisted

# docs-from-site — Build a rich docs site from a product website Goal: turn a product URL into a site that reads like **production documentation a competitor would envy** — not a flat dump of scraped pages. A new visitor who is not yet sold should land on it, understand the value in one screen, and find a clear path deeper. The bar is a foldered, benefit-first site with an FAQ and use-cases, wired into a navigation graph — not "README + one thin page". The intelligence lives in **this skill**, not in a helper tool. Read the site yourself, decide the structure yourself, write the pages yourself, then delegate only *publishing and configuration* to whatever docs platform is connected. --- ## Why the usual crawl fails (read before you fetch) Most modern product sites are JavaScript SPAs. A plain HTTP fetch of `/`, `/docs`, `/features`, `/blog/<slug>` returns an empty HTML shell or a 404 — the real content is behind JS render. If you fetch flat HTML and the `<main>` is thin, you will silently fall back to *inventing* generic content, which is exactly the failure mode that produces a bland site. **So: render first.** Use a real browser (headless / automation) to load each page and read the rendered `<main>` / `<article>`. Fall back to a plain HTTP fetch only for pages that are already static (a classic server-rendered marketing page). Never write a page from a shell you could not actually read — skip it and note why. --- ## Workflow ### 1. Map the site - Fetch `/sitemap.x