This is a reviewable content-site sequence, not a claim that the current CLI has a one-command content-site scaffold. The first board records that gap rather than inventing one.
Build one complete vertical slice
- Start with the verified information-site base.
- Add two objects under
content/articles/<slug>/content.md, each with real front matter and reader body. - Add a page/collection layout that declares an unfiltered
/content/articles/source. - Bind that complete source value to a component with an explicit empty state.
- Confirm one list result and one detail route in every declared locale.
Only after that slice works should you add sorting, a limit, tags, related objects, or an archive. Each one changes selection or route behavior and deserves its own build/browser observation.
The core boundary
The content object remains the source of truth. The list page selects objects; it does not own article copy. The detail route presents one object; it does not duplicate its metadata into site configuration. This lets an object appear in a list, related-reading band, topic page, or future surface without moving the original file.
For exact object shape, see Content objects, metadata, and locales. For the query contract, see Bind content to a page. The board deliberately does not publish a fake “download this full blog” fixture until a clean-agent vertical slice has been run against the target runtime.