Problem

The HTML stdlib is well-designed around CSS selectors (html_query, html_text, html_attr), but html_query always returns Array<HtmlNode> — even when only one element is expected. The common pattern of "get the first matching element" requires an extra array_get + Option unwrap:

let @Array<HtmlNode> = html_query(@HtmlNode.0, "h1");
let @Option<HtmlNode> = array_get(@Array<HtmlNode>.0, 0);
-- now unwrap the Option...

This is the same boilerplate every HTML-scraping program will write.

Proposed additions

-- First CSS selector match — Option<HtmlNode> instead of Array<HtmlNode>
html_query_one(@HtmlNode.0, "h1")     -- Option<HtmlNode>

-- Tag name of an element node (None for text/comment nodes)
html_tag(@HtmlNode.0)                  -- Option<String>

-- Direct children of a node
html_children(@HtmlNode.0)             -- Array<HtmlNode>

Impact

html_query_one covers the dominant scraping pattern — get the page title, get the first paragraph, get the canonical link. html_tag enables dispatching on element type when traversing a node list. html_children enables manual tree traversal for cases where CSS selectors aren't expressive enough.

Implementation

All three are pure Vera prelude functions or trivial extensions of existing host imports. html_query_one is html_query + array_get(_, 0). html_tag and html_children match on the HtmlNode constructor (HtmlElement carries tag + attributes + children; HtmlText/HtmlComment do not).