SEO & AEO

What ships

• Per-page metadata via one getSEOTags() helper: title, description, keywords, robots directives, canonical URL, OpenGraph, and Twitter card.
• JSON-LD structured data (Organization, WebSite, SoftwareApplication, FAQPage, BlogPosting, BreadcrumbList) via renderStructuredData().
• Sitemap and robots.txt generated on every build with per-route priority and lastmod.
• OpenGraph images auto-generated at 1200×630 for every share.
• An AI-crawler allowlist and llms.txt so answer engines can read and cite the site.

Metadata on every page

All SEO tags flow through getSEOTags() in src/lib/seo.tsx. It is already wired into app/layout.tsx, so every route gets sane defaults from config.ts. Each page exports its own metadata to override title, description, and canonical:

import { getSEOTags } from "@/lib/seo";

export const metadata = getSEOTags({
  title: "Pricing — AI Web Launcher",
  description: "One-time price, unlimited sites. 120–160 characters, primary keyword included.",
  canonicalUrlRelative: "/pricing",
  // optional: keywords, openGraph.images, noindex
});

The helper emits an explicit <meta name="robots" content="index, follow"> on every route (with max-image-preview: large for Googlebot), plus a matching OpenGraph object and a summary_large_image Twitter card. To keep a thin or utility page out of the index, pass noindex: true instead of hand-writing a robots tag.

Structured data (JSON-LD)

renderStructuredData() in src/lib/seo.tsx emits a single @graph script per page. It is called once site-wide in app/layout.tsx and per page where richer schema helps. Pass the page type:

The SoftwareApplication offer price is read from config.stripe.plans, so the schema never contradicts your live pricing. Never inject structured data as a raw <script> tag; use the helper so the graph stays valid.

Sitemap and robots.txt

next-sitemap runs on the postbuild script and writes public/sitemap.xml and public/robots.txt. Behavior lives in next-sitemap.config.js:

• Per-route priority and changefreq. Home is 1.0/daily, pricing 0.9/weekly, blog 0.8/daily, docs 0.8/weekly, articles and legal pages lower.
• Accurate lastmod. Blog entries pull their date from src/features/blog/content.tsx; other routes fall back to git history.
• Exclusions. Image and icon routes, /api/*, and /llms.txt are kept out of the sitemap. Add a private page to the exclude array to hide it.

module.exports = {
  siteUrl: process.env.SITE_URL || "https://your-domain.com",
  generateRobotsTxt: true,
  exclude: ["/api/*", "/llms.txt", "/*.png", "/opengraph-image", /* … */],
  transform: async (config, loc) => {
    // per-route priority + changefreq + lastmod
  },
};

OpenGraph images

app/opengraph-image.tsx generates a 1200×630 share image at the extension-less route /opengraph-image. getSEOTags() points both the OpenGraph and Twitter tags at that route, so every share renders an image with no manual asset work. Blog posts generate their own image from the article definition.

AEO: getting cited in AI answers

An LLM composes its answer from many sources, so ranking third is not the bar anymore. Any page that adds one specific, useful detail can be pulled into the answer. The boilerplate ships the plumbing that lets answer engines read and quote you.

llms.txt

src/app/llms.txt/route.ts is a force-static route that serves /llms.txt, a curated, JavaScript-free Markdown map of your site for language models (spec: llmstxt.org). It reads name, description, domain, and support email from config.ts, so it updates when you rebrand. When you add a high-intent page (pricing, a comparison, a category page), add a line for it in the route; it is the AEO counterpart to sitemap.xml and is deliberately excluded from the sitemap so it is not treated as an indexable HTML page.

AI-crawler allowlist

next-sitemap.config.js (mirrored in public/robots.txt) explicitly allows the 2026 AI-crawler set. Every bot you block is an answer engine you disappear from, so the list stays open unless a content-protection decision says otherwise:

OpenAI      GPTBot · OAI-SearchBot · ChatGPT-User
Anthropic   ClaudeBot · Claude-SearchBot · anthropic-ai
Google      Google-Extended (Gemini / AI Overviews)
Others      PerplexityBot · Applebot-Extended · Amazonbot · CCBot · cohere-ai · Bytespider

The -SearchBot and ChatGPT-User agents put you in the AI answer; the rest cover training and indexing. To add a bot, append { userAgent: "NewBot", allow: "/" } to the policies array.

FAQs that get quoted

LLMs lean on question-and-answer format and often cite FAQ blocks verbatim. The homepage FAQ lives in a single liveLandingFaqItems array in src/lib/seo.tsx. It is rendered visibly on the page and emitted as FAQPage JSON-LD from the same source, so the two can never drift. Two rules make it work:

• Phrase each question in the exact words a user types into a chatbot, not product-FAQ phrasing.
• Keep each answer standalone in 2 to 4 sentences; the model may quote it without the rest of the page.

Product-FAQ (weak):   "Do I get updates?"
Chatbot (strong):     "What is the best Next.js boilerplate for AI-built sites?"
                      "Is there a <competitor> alternative that does <differentiator>?"

Pages the schema is built for

These are the page types answer engines cite most. The boilerplate ships the schema helpers (FAQPage, BreadcrumbList) they need; you write the pages:

• Category pages. A "best X for Y" landing page that anchors the exact phrase you want to be recommended for, with a chatbot-phrased FAQ at the bottom.
• Honest comparison pages. A /vs/<competitor> page that concedes where the competitor genuinely wins. The honesty is the citation mechanism; models have read everyone's marketing and can tell an evaluation from a sales page. Verify competitor pricing at the source on the day you write it.
• Use-case pages. One job-to-be-done per page, specific and citable.