---
title: Add llms.txt and Markdown endpoints to Fumadocs
description: Reuse a Fumadocs MDX content collection to generate AI-readable discovery and page endpoints in a Next.js App Router deployment.
canonical_url: https://next-ai-ready.vercel.app/en/docs/guides/fumadocs-ai-ready
url: https://next-ai-ready.vercel.app/en/docs/guides/fumadocs-ai-ready
last_updated: 2026-08-02
updated: 2026-08-02
author: next-ai-ready team
summary: Reuse a Fumadocs MDX content collection to generate AI-readable discovery and page endpoints in a Next.js App Router deployment.
topics: [fumadocs, nextjs, llms.txt, mdx]
---

# Add llms.txt and Markdown endpoints to Fumadocs

[Fumadocs](https://www.fumadocs.dev/docs) provides the documentation UI, Source API, and search.
`next-ai-ready` can read the same MDX collection to produce `/llms.txt`, page Markdown, semantic page
discovery, and optional MCP resources without changing those Fumadocs responsibilities.

This integration targets Fumadocs on Next.js App Router with a server deployment. It does not support
a fully static `output: "export"` build because the machine-facing route handlers run at request time.

## 1. Install and initialize

From the Fumadocs Next.js application directory:

```bash
pnpm add next-ai-ready zod@^4
pnpm exec next-ai-ready init
```

Keep `source.config.ts`, the generated Fumadocs source module, page layouts, and search routes intact.
The initializer adds separate AI-facing handlers and configuration.

## 2. Reuse the Fumadocs content collection

Fumadocs MDX commonly stores documentation under `content/docs`. Point next-ai-ready at the same
source files:

```js
import { defineConfig } from "next-ai-ready"

export default defineConfig({
  site: {
    name: "Acme Docs",
    baseUrl: "https://docs.example.com",
    description: "Documentation for Acme developers.",
  },
  content: ["content/docs/**/*.{md,mdx}"],
})
```

If the collection lives under `src/content/docs`, use that path instead. Existing `title`,
`description`, `date`, and `author` frontmatter are reused. Add `updatedAt` only when the date is
truthful and maintained.

## 3. Wrap the Next.js config

Fumadocs does not require a competing Next.js wrapper for its Source API. Preserve the project's
existing options and wrap them with `withAiReady()`:

```js
import { withAiReady } from "next-ai-ready/config"

const nextConfig = {
  // Existing Fumadocs and Next.js options stay here.
}

export default withAiReady({ agentReadable: true })(nextConfig)
```

If another plugin already wraps the config, apply `withAiReady()` to that plugin's final result.

## 4. Build and verify

```bash
pnpm exec next-ai-ready build
pnpm exec next-ai-ready doctor --score
pnpm build
```

After deployment:

```bash
curl -I https://docs.example.com/llms.txt
curl -I https://docs.example.com/docs/getting-started.md
pnpm exec next-ai-ready audit https://docs.example.com/docs/getting-started --version 3
```

Confirm that `/llms.txt` lists the intended Fumadocs pages, Markdown URLs use production canonicals,
and browser requests still receive the normal Fumadocs UI.

## Keep search responsibilities separate

Fumadocs search serves people inside the documentation UI. next-ai-ready page discovery serves AI
clients through Markdown, HTTP resources, and optional MCP. They may index the same content, but one
does not need to replace the other.