Your website now has a second reader. llms.txt and .md files, explained for owners.

To answer one question, America.gov has to read government pages written for people. Menus, banners, cookie notices, sidebars, PDFs. Somewhere in the middle is the three sentences that answer the question.

Every AI assistant does the same thing with your website. It is your second reader, and it reads very differently from the first.

What the second reader wants

People want design. Machines want the facts, in order, without the decoration. Four things help most.

1. llms.txt. A plain text file at yoursite.com/llms.txt that tells an AI, in a few hundred words, what your business is, what you sell, and which pages matter. Think of it as the note you would leave for a smart new hire on their first day.

2. Markdown versions of pages. A .md file is the text of a page without the page. Headings, paragraphs, lists, links. An agent can read it in one pass with nothing in the way. Some sites publish a .md twin of every important page at the same address with .md on the end.

3. Structured facts. A small block of JSON-LD on your pages that states your name, address, hours, prices and services in a format machines already understand. Written once, read by every assistant.

4. Content that loads without JavaScript. If your text only appears after scripts run, many agents see an empty page.

Try it on this post

Every post in this series has a markdown twin. Add .md to the address:

https://orgtp.com/blog/your-website-has-a-second-reader.md

That is what an agent reads. No menu, no footer, just the argument.

Our site also serves orgtp.com/llms.txt, and a machine-readable file at /.well-known/mcp.json that lists the tools an agent can call on the page.

What we have seen move the score

Agent Ready scores a site out of 100 across four pillars: can an agent reach it, understand it, act on it, and finish the job.

Two of our own sites went from a D to an A after we made these changes:

• orgtp.com went from 51 to 90.
• ijoin.ai went from 41 to 92.

The biggest gains were not in reading. They were in acting. The heaviest single check in the scan is whether a page tells agents what they can do there, through a new browser standard called WebMCP.

For perspective, stripe.com scored 64 when we scanned it. It was perfect at letting agents in, and scored 9 out of 100 on letting them act. If Stripe has room to grow here, everyone does.

The government is doing the same work

Federal agencies are not just putting a chatbot on top. They are publishing their data so any AI can read it directly, through something called MCP. That is the next post.

What to do this week

1. Visit yoursite.com/llms.txt. If it is a 404, write one. Five paragraphs is plenty.
2. Pick your three most important pages and check that their text shows up with JavaScript turned off.
3. Run a free scan at agentready.sneeze.it and fix the top finding first.

Frequently asked questions

What is llms.txt?

llms.txt is a plain text file at the root of a website that tells AI systems what the site is about and which pages matter most, in a format that is easy for a language model to read.

Why publish markdown versions of web pages?

A markdown version gives AI agents the text of a page without navigation, scripts and layout, so they read the content faster and more accurately. Many sites publish it at the same address with .md added.

What is WebMCP?

WebMCP is an emerging browser standard that lets a web page declare tools an AI agent can call, such as searching a catalog or booking an appointment, so the agent does not have to guess by clicking.

Sources

• Agent Ready, Sneeze It
• orgtp.com llms.txt
• The llms.txt proposal

Ask your AI who owns what

OTP is readable from any MCP client. In Claude Desktop, Cursor or any MCP client, add this block:

"otp": {
  "command": "npx",
  "args": ["-y", "@orgtp/mcp-server"]
}

Restart the client. Then ask: "Use OTP to show me Sneeze It's chart. Who owns each seat, and what number is each one measured on?"

Start free at orgtp.com. Every seat is free, for people and agents. You pay only for the AI you use.

The series

1. America.gov in one page: what the government's AI front door teaches every business about its knowledge problem
2. The government has 29,000 websites. Your company has the same problem in Google Drive.
3. America.gov solved the easiest 5%. The other 95% is who owns the answer.
4. Every answer America.gov gives links to its source. Your company's answers should too.
5. America.gov is built on AI. Its robots.txt blocks AI agents. Check yours.
6. Your website now has a second reader. llms.txt and .md files, explained for owners.
7. The quiet part of the government's AI push is MCP, and it matters more than the chatbot.
8. America.gov changed what it would answer during its own launch. Decide what your AI won't answer before yours.
9. America.gov answers now and acts in 2027. That is the right order for your agents too.
10. One front door only works if someone owns every room behind it.

All ten on one page: What America.gov teaches business about its knowledge problem.

Series: What America.gov teaches business about its knowledge problem. Part 6 of 10. Read this post as markdown: /blog/your-website-has-a-second-reader.md.