Search inside templates
How Everything surfaces related public articles alongside an individual page.
Everything can show related reading alongside an article without requiring each article to maintain a manual list of every possible connection.
Two kinds of links
Curated links come from the article's own content. They express relationships chosen by the author.
Search-generated links are discovered when a page is displayed. The selected page supplies a short description, and a bounded lexical search finds other eligible public articles. The current article itself is excluded from the suggestions.
The related panel is separate from the main article. On wide screens it can appear in a sidebar; on smaller screens it follows the reading column. It loads only when needed so that opening a page does not have to wait for every optional search result.
Why lexical search?
The lightweight related-reading path uses lexical retrieval rather than running a full semantic/auto search for every article view. Full Search is available when the reader or agent wants broader discovery.
Results are snapshots and can be incomplete or temporarily stale. A page's inclusion still depends on public eligibility; search is not an alternative way to access drafts or private assets.
Templates remain declarative
The related panel is part of a stored layout interpreted by Everything. Template definitions describe presentation and bounded public discovery; they are not executable programs. How pages become websites explains the rendering process.
Agents do not need to render the sidebar at all: public MCP tools such as get_related_pages, list_pages and search_pages expose the same discovery capabilities directly.
For more, read Search and browse, Caching and freshness and Web, JSON and MCP.