NuxtSEO Pro
Core Concepts

Markdown Conversion

Last updated by Harlan Wilton in doc: link.

Nuxt AI Ready uses mdream to convert HTML pages to markdown during prerendering and runtime.

Conversion Flow

Build-Time (Prerendering)

During nuxi generate or nuxi build --prerender:

  1. HTML processed via mdream extraction plugin (title, description, headings)
  2. Converted to markdown and chunked
  3. Written to bulk TOON at /llms-full.toon (chunk-level) and /llms.toon (page-level)
  4. Full markdown served at .md routes
// Chunking config (reference)
const chunks = htmlToMarkdownSplitChunksStream(html, {
  headersToSplitOn: [TagIdMap.h1, TagIdMap.h2, TagIdMap.h3],
  chunkSize: 256,
  lengthFunction: estimateTokenCount,
})

Runtime

Requests with .md extension or Accept: text/markdown header trigger on-the-fly conversion with cache headers.

curl https://example.com/about.md
curl -H "Accept: text/markdown" https://example.com/about

Configuration

export default defineNuxtConfig({
  aiReady: {
    mdreamOptions: {
      preset: 'minimal', // Use minimal preset
    },
    markdownCacheHeaders: {
      maxAge: 3600, // Cache for 1 hour
      swr: true, // Enable stale-while-revalidate
    },
  },
})

Hooks

ai-ready:mdreamConfig

// Dynamic config based on route
nitroApp.hooks.hook('ai-ready:mdreamConfig', (options) => {
  if (options.origin?.includes('/blog/'))
    options.ignoreElements = [...(options.ignoreElements || []), '.author-bio']
})

// Full docs: /docs/ai-ready/nitro-api/nitro-hooks#ai-ready-mdreamconfig

ai-ready:markdown

// Post-process markdown output
nitroApp.hooks.hook('ai-ready:markdown', (context) => {
  context.markdown = `---\ntitle: ${context.title}\n---\n\n${context.markdown}`
})

// Full docs: /docs/ai-ready/nitro-api/nitro-hooks#ai-ready-markdown

Learn More

Edit this page
Markdown For LLMs
Did this page help you?

llms.txt Generation

What is llms.txt and how we can modify it with Nuxt AI Ready.

RAG Integration

Build RAG pipelines with vector embeddings using the ai-ready:chunk hook for real-time chunk processing.