Markdown on every public page
An AI agent that reads a web page wants the text, not the navigation, scripts and styling around it. Until November 19, Namegarage gave agents that text in two places. Blog posts had Markdown copies at their own .md addresses, and the homepage answered a request for Markdown. Every other page sent only HTML.
Release 2026-11-19-markdown-on-every-public-page makes every public page answer a request that sends Accept: text/markdown with Markdown. It is live on namegarage.com.
Why we do not write a Markdown file for each page
The obvious design is a .md file next to each page, as the blog has. For a domain page, that file would soon be wrong. A domain page's price and availability are set when someone requests the page: a weekly deal can start or end, and a name can sell. A file written when the site was built would keep the old price until the next deployment. An agent would then read one price in the Markdown copy while the page showed another.
So we do not write those files. Cloudflare converts the page to Markdown when the request arrives, from the same final HTML that a browser receives. The Markdown shows the price that the page shows.
Our guide for agents, /llms.txt, describes the site and lists its documents. It does not carry the pages' content, so it cannot replace page Markdown.
How to ask for Markdown
Send the same request to the same address with one extra header:
curl -H 'Accept: text/markdown' https://namegarage.com/about/
This works on the homepage, domain pages, catalog pages, Deals, the blog, the podcast, the trust center, the legal pages, the About, Brand, Sales, Contact, Refer and Partners pages, and the developer documentation. A browser keeps receiving HTML. A HEAD request returns the same headers without a body.
The Markdown starts with the page's title and description and leaves out the navigation, header and footer. Structured data in the page stays as a fenced JSON block.
A program that needs fields rather than prose should still use /domains.json or the read-only search API.
What changes in the response
Each public HTML page states that its body depends on the requested format, and it states the site's usage policy:
Vary: Accepttells caches to keep the HTML and Markdown versions apart.Content-Signal: search=yes, ai-input=yes, ai-train=yesrepeats the policy inrobots.txt. The Markdown version keeps it.
The Markdown response has no ETag or Last-Modified header, so a conditional request returns the full body every time. Blog posts keep their .md copies, and the developer documentation and agreements keep their own Markdown files.
One list of public pages
We already keep one reviewed list of public pages that ordinary HTTP clients may read without a browser check. Markdown uses that same list, so a page that a script can read can also be read as Markdown. Six information pages that were missing from the list joined it in this release: About, Brand, Sales, Contact, Refer and Partners. Checkout, sign-in, partner workspaces, agreements and shared shortlists stay outside the list, and Cloudflare does not convert them.
Checked on the live site
After the November 19 deployment, we checked the live site with curl and Python's urllib, with no browser and no sign-in. We requested the homepage, two domain pages, a catalog page, the terms, the documentation, the blog, two blog posts and the six added pages. Without the header, each page answered GET and HEAD requests with HTML; with it, each answered with Markdown. Every response carried Content-Signal and Vary: Accept. The six added pages and /sales.json answered urllib without a browser check. Checkout, a shared shortlist, an agreement and an API route returned no Markdown.

Comments
Comments are reviewed before publication. The Author badge marks replies from the author.
Your test session has expired. Stop testing to use comments and sign-in normally, or open a fresh test link to keep testing.
Stopping testing reloads this page in normal mode.
Read approved comments on the discussion page.
Open the discussion page