---
title: "agent-ready.website"
date: 2004-03-19
modified: 2038-01-19
author: "agent-ready.website"
url: "https://agent-ready.website/"
markdown_url: "https://agent-ready.website/index.md"
type: page
description: "A single Markdown file that scores 100 on agent readiness, with nothing else."
---

# Agent-Ready, Finally.

And it scores 100.

## What more could you ask for?

Perhaps you build agent-facing sites and believe yours is especially distinctive. You may think your 13-megabyte JSON-LD graph will earn you a badge in the corner of your `llms.txt`. Perhaps you imagine that a 40-file `/.well-known/` directory and a `<link rel="alternate">` in every header will make the crawlers rejoice. I'm afraid not. Allow me to describe the ideal document:

- It is one file, and it is named `index.md`.
- It contains no markup that a human would need to read around.
- It is served as `text/markdown`, with the charset declared.
- It makes no promises that a second file would have to keep.
- It is easy to read and communicates its message clearly, assuming it has one.

## Here is the point

You may be overengineering. Take a look: it is a Markdown file. Why generate a `.md` twin for every page when an agent only ever reads the front door? You may have spent a week on a schema-validated feed and added 80 kilobytes to your build, yet no one has ever requested `/about.md`.

You may not have realized it, but this is what a good agent-ready site can be. Here's why.

## It is lightweight

This entire document weighs less than the `llms-full.txt` on your last project. Did you really need a 100-kilobyte JSON-LD block just to say that a page is a page? Did you need seven manifests just to say "Hi"?

## It is parseable

You may think you need a dozen structured-data formats for an agent to understand a site. Parseability simply means that a file is valid Markdown. This one parses whether the reader is a language model or a `grep` command.

## It works

You can read it, provided you are able to read. It has headings, so tools and agents can understand its structure. That is what headings are for.

The content is right there in the file. Your site may have a robots.txt with an essay of crawler etiquette, a llms.txt with a table of contents for the table of contents, and a 404 page that returns the sitemap. All of it could have been a single heading and a few lines.

As for compatibility, try serving this file as `application/octet-stream` and see what happens. I dare you.

## This is a Markdown. Take a look. Perhaps you have never seen one quite like it.

Just as someone who has never seen a plain text file may not know what documentation is meant to look like, you may not know what a website is meant to be. Look at it. It scores 100.

### Yes, this is satire

I am not actually suggesting that your site should be a single Markdown file. My point is that many of the problems agents encounter are **ones we create ourselves**. Documents are not inherently unreadable. We are the ones who sometimes make them otherwise.

> "Making the simple complicated is commonplace. Making the complicated simple—awesomely simple—that's creativity."
> —Charles Mingus, statement in *Mainliner Magazine*, July 1977

*Inspired by [the original](https://motherfuckingwebsite.com)*