# Accept: text/markdown

**Serve Markdown to agents with content negotiation.** Your site already has the content. Serving a Markdown variant lets agent clients skip nav/scripts/layout markup and read the content directly.

## Three benefits

- **Tokens — a fraction of the bytes.** Markdown drops nav, styles, scripts, and layout wrappers. Agents spend context on your prose, not your DOM.
- **Retrieval — higher signal-to-noise.** No ads, related-content rails, or modal overlays muddying the text a RAG pipeline has to embed.
- **Latency — faster first token.** Less to fetch, less to parse, less to stuff into the context window before the model starts thinking.

## Check your site

The HTML version of this page has two interactive tools:

- **Readiness check** — scores a URL against four checks: serves Markdown for `Accept: text/markdown`, sets `Vary: Accept`, rejects unsupported types with `406`, and honors q-values.
- **Accept tester** — makes a request from our edge with a given `Accept` header and shows the raw response headers and body.

You can run the same checks yourself with curl:

```bash
curl -sI -H "Accept: text/markdown" https://yoursite.com/article
curl -s  -H "Accept: text/markdown" https://yoursite.com/article
```

Look for `Content-Type: text/markdown; charset=utf-8` and `Vary: Accept` on the response.

## Where to next

- [Quick start](https://acceptmarkdown.com/start) — the smallest working setup
- [Guides](https://acceptmarkdown.com/guides) — fundamentals, correctness, generating
- [Recipes](https://acceptmarkdown.com/recipes) — Nginx, Caddy, WordPress, Laravel, Rails, Cloudflare Workers, Next.js, Astro, Apache, SvelteKit, Nuxt/Nitro, Express, Django
- [Status](https://acceptmarkdown.com/status) — which AI agents send `Accept: text/markdown` when their browse or fetch tools hit a URL
- [Reference](https://acceptmarkdown.com/reference) — RFC 7763, RFC 9110, specs, adjacent work