Agents that ask for Markdown get your page's Markdown copy at the same address

More and more sites ship a Markdown copy of each page beside the HTML, for AI agents and other tools that read text better than markup. Until now an agent had to know your site’s naming convention to find that copy. Now it can ask for the page’s usual address and say it prefers Markdown, and the edge hands it the copy.

Add this line anywhere in your _redirects file:

# ee:markdown on

The line is a comment, so the same file still works unchanged elsewhere.

Who gets Markdown

A request gets the Markdown copy only when its Accept header ranks text/markdown above text/html. A tie goes to HTML, so browsers, and requests that accept */*, always get your HTML page as before.

Where the copy is looked for

• /cli/ → /cli.md, then /cli/index.md
• / → /index.md
• /about.html and /about → /about.md

If a page has no Markdown copy, nothing changes for it.

How it behaves

• Your own rules come first. A path you redirect, or rewrite to another file, is left alone, so a single-page-app catch-all doesn’t hand out your home page’s copy for every route. Clean URLs still work.
• The copy is served as the file it is, with its own content type and the _headers rules for the .md path, such as a noindex rule for /*.md.
• It takes the page’s cache policy, so a new deploy shows at once.
• Both versions carry Vary: Accept, and the HTML page also carries Link: </cli.md>; rel="alternate"; type="text/markdown", so a tool reading the HTML can find the copy. Pages without a copy get neither header, and any Link or Vary rules of your own stay alongside ours.
• A Markdown response counts as a normal request in your usage.

Limits

• Level 0 sites only, like everything in _redirects.
• There’s no CLI flag for it yet. Add the line to _redirects yourself.

What you need to do

Nothing, unless you want it. If your site already builds .md copies of its pages, add the line above to _redirects and deploy.