"The official s&box docs are machine-readable via llms.txt and .md URLs"

Scraping or fetching the HTML of an s&box documentation page returns an empty shell with no readable content. The page renders fine in a browser but the server-sent HTML contains only the Blazor app bootstrap, not the documentation text.

The s&box documentation site is a client-rendered Blazor app. All content is loaded dynamically on the client side after the initial page load. The server sends only the application shell, so any tool that reads the raw HTTP response (curl, fetch, scrapers) gets no documentation content.

Use the built-in machine-readable endpoints instead of scraping:

• Full index: https://sbox.game/llms.txt provides a complete documentation index (advertised in robots.txt via the Llms-Txt: directive).
• Raw markdown per page: append .md to any documentation URL to get the raw markdown source. For example, /dev/doc/scene/components becomes /dev/doc/scene/components.md.

Scope limits (verified): the .md trick covers /dev/doc/* pages only. News posts do not serve .md; for news, use the RSS feed at https://sbox.game/rss/news, which is complete, dated, and carries full HTML in each item description. There is no sitemap.xml.

Important: treat docs as supporting evidence for engine behavior claims. The docs describe intent; a live editor check remains the verification authority for how things actually behave.

The documentation content is stored as markdown on the server and served through a dedicated endpoint (the .md suffix route) separate from the Blazor rendering pipeline. The llms.txt index follows the emerging standard for making sites machine-readable to language models, providing structured access without needing to execute client-side JavaScript.