How to Add llms.txt to Your Site (Astro, Next.js, WordPress)

By Hunter Spence · 2026-06-11 · 8 min read

llms.txt is a plain Markdown file served at your domain root that gives AI systems a curated index of your site: what it is, which pages matter, and how to interpret them. ChatGPT, Claude, and Perplexity all parse it when present. Deployment takes ten minutes on any stack; this guide covers the format once, then the exact steps for Astro, Next.js, and WordPress.

The format (90 seconds)

A valid llms.txt needs exactly three structural elements, in this order:

# Your Site Name

> One sentence describing what the site is and who it serves.

https://yourdomain.com/

## Main Sections
- [Page Title](https://yourdomain.com/page): one-line description of the page
- [Another Page](https://yourdomain.com/another): what an AI should know it covers

## Docs
- [Getting Started](https://yourdomain.com/docs/start): installation and first run

• The # H1 site name must be the first content line.
• The > blockquote is your one-sentence elevator description — models quote it.
• Sections (##) group - [Title](URL): description entries. Absolute https URLs only.
• List pages you'd want an AI to cite — not every URL. Quality beats coverage; ~10-150 entries is the useful range.

Don't hand-roll it if you have a sitemap: the free generator builds a spec-valid file from your own sitemap, so every URL is verified to exist.

How do I add llms.txt in Astro?

Astro serves anything in public/ verbatim from the site root. Create public/llms.txt, paste your content, deploy. That's the whole implementation:

your-project/
├── public/
│   ├── llms.txt        ← here
│   └── robots.txt
└── src/…

Astro won't process or rename it; it ships byte-for-byte at /llms.txt with a sensible text/plain content type. If you generate pages from a content collection, you can also render it dynamically with an endpoint at src/pages/llms.txt.ts that walks your collection and returns the Markdown — useful when your page list changes weekly.

How do I add llms.txt in Next.js?

Static (App or Pages router): identical idea — drop the file in public/llms.txt. Next serves public/ from the root.

Dynamic (App router): create a route handler so the file always reflects current content:

// app/llms.txt/route.ts
export async function GET() {
  const body = [
    '# Acme',
    '',
    '> Acme makes billing infrastructure for veterinary clinics.',
    '',
    'https://acme.com/',
    '',
    '## Product',
    '- [Pricing](https://acme.com/pricing): plans and per-seat pricing',
    // …generate entries from your CMS or route manifest
  ].join('\n');
  return new Response(body, {
    headers: { 'Content-Type': 'text/plain; charset=utf-8' },
  });
}

Avoid middleware rewrites that might intercept /llms.txt — check it returns 200 in production, not your SPA shell.

How do I add llms.txt in WordPress?

Three options, best first:

1. Upload to the web root. Via SFTP or your host's file manager, place llms.txt next to wp-config.php. Survives theme changes; bypasses WordPress entirely.
2. Plugin. Several maintained plugins (search the directory for "llms.txt") generate the file from your published pages and keep it updated on publish — the right choice if you don't have file access.
3. functions.php rewrite for the comfortable: register a rewrite for llms.txt and emit the Markdown from a template. More moving parts; only worth it for heavily dynamic sites.

Watch for caching/security plugins and CDN rules that block unknown root files — after deploying, fetch https://yoursite.com/llms.txt from an incognito window and confirm a 200 with your content.

The four mistakes that suppress citations

1. Wrong location. It must be at the root. /about/llms.txt doesn't exist to anyone.
2. Listing dead or noindexed URLs. Models treat inconsistency between llms.txt, sitemap, and live pages as unreliability. Our audit cross-checks every entry against your sitemap for exactly this reason.
3. Serving HTML. A 404 page returning 200, or a pretty "viewer" — content-type must be plain text/Markdown.
4. Set-and-forget. A file describing your 2024 site map misleads about your 2026 one. Regenerate when your information architecture changes (the $49/mo re-audit alerts on drift).

Validate, then verify live

Paste your file into the validator (11 spec checks, instant), deploy, then run the full audit — it fetches the live file, scores structure, and verifies every URL against your sitemap. Green on checks 07 and 08 means the index layer of your GEO stack is done.