feat(0.7.0): @cdr-kit/forms + fumadocs content infra + real Aeneid everywhere

apps/site/.gitignore

@@ -18,3 +18,7 @@ yarn-debug.log*
 .DS_Store
 .env.local
 .env*.local
+.source
+
+# T1 migration temp backup of TSX docs pages
+_docs_old_tsx_backup/

apps/site/app/api/search/route.ts

@@ -0,0 +1,4 @@
+import { source } from "@/lib/source";
+import { createFromSource } from "fumadocs-core/search/server";
+
+export const { GET } = createFromSource(source);

apps/site/app/docs/[[...slug]]/page.tsx

@@ -0,0 +1,95 @@
+import { notFound } from "next/navigation";
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
+import type { Metadata } from "next";
+import type { ComponentType } from "react";
+import { source } from "@/lib/source";
+import { DocPage } from "@/components/docs/doc-page";
+import { CopyPageButton } from "@/components/docs/copy-page-button";
+import { getMDXComponents } from "@/lib/mdx-components";
+
+type Props = { params: Promise<{ slug?: string[] }> };
+
+// The frontmatter schema (defined in source.config.ts) doesn't flow into source.getPage().data
+// as typed fields — fumadocs surfaces them as `unknown` by default. This interface is the
+// runtime shape we know exists.
+interface DocFrontmatter {
+  title: string;
+  description?: string;
+  importLine?: string;
+  breadcrumb?: string[];
+  prev?: { href: string; label: string };
+  next?: { href: string; label: string };
+  body: ComponentType<{ components: ReturnType<typeof getMDXComponents> }>;
+  toc: Array<{ url: string; title: unknown; depth: number }>;
+}
+
+export async function generateMetadata({ params }: Props): Promise<Metadata> {
+  const { slug } = await params;
+  const page = source.getPage(slug);
+  if (!page) return {};
+  const data = page.data as unknown as DocFrontmatter;
+  return {
+    title: `${data.title} — cdr-kit`,
+    description: data.description,
+  };
+}
+
+export async function generateStaticParams() {
+  return source.generateParams();
+}
+
+export default async function Page({ params }: Props) {
+  const { slug } = await params;
+  const page = source.getPage(slug);
+  if (!page) notFound();
+
+  const data = page.data as unknown as DocFrontmatter;
+  const MDX = data.body;
+
+  // Map fumadocs TOC ({ url: "#id", title, depth }) → DocTocItem ({ id, label }).
+  // Only h2/h3 land in the right rail to match the existing scrollspy density.
+  const tocItems = data.toc
+    .filter((node) => node.depth <= 3)
+    .map((node) => ({
+      id: node.url.replace(/^#/, ""),
+      label: typeof node.title === "string" ? node.title : "",
+    }));
+
+  // Raw MDX for the per-page Copy button. Server-read once; passed as a string
+  // to the client component so there's no extra fetch round-trip. The MDX file
+  // path is derived from the URL: /docs/components/x → content/docs/components/x.mdx;
+  // /docs → content/docs/index.mdx.
+  let rawMarkdown = "";
+  try {
+    const rel = page.url.replace(/^\/docs\/?/, "") || "index";
+    const mdxPath = join(process.cwd(), "content/docs", `${rel}.mdx`);
+    rawMarkdown = readFileSync(mdxPath, "utf-8");
+  } catch (err) {
+    // Surface but don't crash — the page itself is already rendered via fumadocs
+    // page.data.body; the Copy button just degrades to an empty payload.
+    console.warn(`[docs] Failed to read raw MDX for ${page.url}:`, (err as Error).message);
+  }
+
+  return (
+    <DocPage
+      data={{
+        breadcrumb: data.breadcrumb ?? [],
+        title: data.title,
+        badges: <CopyPageButton markdown={rawMarkdown} />,
+        lede: data.description ?? "",
+        importLine: data.importLine,
+        prev: data.prev,
+        next: data.next,
+        sections: [
+          {
+            id: "content",
+            title: "",
+            content: <MDX components={getMDXComponents()} />,
+          },
+        ],
+      }}
+      tocItems={tocItems}
+    />
+  );
+}