feat(docs): split llms.txt into spec-compliant index and full dump (#2283)

Author: asblanco

packages/stacks-docs/.eleventy.js

@@ -18,8 +18,12 @@ module.exports = function(eleventyConfig) {
   eleventyConfig.addPlugin(iconPlugin);
   eleventyConfig.addPlugin(headerPlugin);
   eleventyConfig.addPlugin(tipPlugin);
+  // Produces /llms-full.txt — a content dump for tools that want the full
+  // markdown of every page in one file. The spec-compliant index lives at
+  // /llms.txt and is rendered by llms.11ty.js.
   eleventyConfig.addPlugin(llmsTxtPlugin, {
     siteUrl: 'https://v2.stackoverflow.design',
+    outputPath: 'llms-full.txt',
     collections: ['base', 'components', 'develop', 'foundation'],
     additionalMetadata: ['description'],
     normalizeWhitespace: true,

packages/stacks-docs/llms.11ty.js

@@ -0,0 +1,60 @@
+// Renders the spec-compliant llms.txt index at /llms.txt
+// (see https://llmstxt.org). The full content dump lives at /llms-full.txt,
+// generated by eleventy-plugin-llms-txt — see .eleventy.js.
+
+const SECTIONS = [
+  { tag: "base", heading: "Base utilities" },
+  { tag: "components", heading: "Components" },
+  { tag: "develop", heading: "Develop" },
+  { tag: "foundation", heading: "Foundation" },
+];
+
+// Descriptions may contain inline HTML for the rendered page; strip it for
+// llms.txt. Loop until stable so unclosed/nested-looking tags can't reintroduce
+// the pattern (CodeQL js/incomplete-multi-character-sanitization).
+function stripHtml(input) {
+  let previous;
+  let output = input;
+  do {
+    previous = output;
+    output = output.replace(/<[^>]+>/g, "");
+  } while (output !== previous);
+  return output;
+}
+
+function pageLink(page, siteUrl) {
+  const title = page.data.title;
+  const description = stripHtml(page.data.description || "")
+    .replace(/\s+/g, " ")
+    .trim();
+  const suffix = description ? `: ${description}` : "";
+  return `- [${title}](${siteUrl}${page.url})${suffix}`;
+}
+
+module.exports = class {
+  data() {
+    return {
+      permalink: "/llms.txt",
+      eleventyExcludeFromCollections: true,
+    };
+  }
+
+  render({ collections, site }) {
+    const sections = SECTIONS.map(({ tag, heading }) => {
+      const pages = (collections[tag] || [])
+        .slice()
+        .sort((a, b) =>
+          (a.data.title || "").localeCompare(b.data.title || "")
+        );
+      const lines = pages.map((p) => pageLink(p, site.url));
+      return `## ${heading}\n\n${lines.join("\n")}`;
+    });
+
+    return `# ${site.title}
+
+> ${site.description}
+
+${sections.join("\n\n")}
+`;
+  }
+};