Give structured data its own page It outgrew a section inside SEO: there is a settings screen, a page type select with nine types, an FAQ editor, breadcrumb modes, a block hook and a partial for routes Atelier does not own. Leads with what a real page emits, because the graph is easier to understand from one example than from a list of node types. Ends with how to check it, and which schemas stopped earning rich results, since half the advice online predates those changes.
Document structured data, both ways in A block can describe itself with structuredData(), and the shipped FAQ block does. But a site's blocks are usually written by whoever installed this, so typing schema on the page is the normal path rather than a fallback, and the docs now say so. Also removes the 'no JSON-LD' lines from four pages, which stopped being true, and replaces them with what is genuinely still missing: Review, opening hours, and nodes from a host app's own models.
Rework the wiki: six new pages, and stop the duplication Everything past installation lived in two half-pages and a lot of it was nowhere. Adds Blocks, Design tokens, Bilingual and RTL, Publishing, Configuration and Troubleshooting, each covering what was previously scattered or missing entirely. Troubleshooting is the one worth having: almost every failure mode in this package is silent, so it lists them in the order they catch people rather than by subsystem. Usage stops repeating the block-writing guide and points at it. How it works gains the pieces built since it was written: the revisions and redirects tables, layout resolution, and why tokens are stored as references. Home becomes a real index, and the sidebar is grouped rather than a flat list.
Document layouts A page can now be wrapped in whichever shell suits it, so document registering them, writing one, and the four things that bite: both partials, the data-atelier-canvas hook, guarding $page, and what happens to a page whose layout was deleted.
Add an SEO page, with a worked example for non-Atelier URLs Everything the SEO layer does was scattered across Installation and nothing documented sitemap sources, which is the part a real client site needs: a blog or a services resource with its own model and routes has URLs that belong in the same sitemap. Includes the whole example end to end, the three shapes a source can return, when to move from a closure to an invokable class, and the four behaviours that surprise people.
Updated Home (markdown)
Document setup, usage and an agent quickstart
Initial Home page