Case studyHow this site is built

Built to be read by machines.

Most websites are written for people and guessed at by AI. This one is written for both. Here is every layer, with a live example you can open.

01
Who I am, as data
Structured data tells search engines and assistants exactly who I am, what I run and since when, so nobody has to infer it.
"@type": "Person", "jobTitle": ["Founder & CEO, Skill Reality", ...]See it in Google’s test
02
A front door for AI
An llms.txt file: a plain summary of the site written for language models, pointing to everything else they need.
/llms.txtOpen it
03
Every page as text
The whole site as one Markdown file, plus a Markdown copy of each page. Clean text, no layout to strip.
/llms-full.txt · /index.mdOpen it
04
The facts, as a file
Roles, dates, companies and writing in one JSON file, so every assistant gives the same answers.
/facts.jsonOpen it
05
A server assistants can query
An MCP server lets AI tools search my writing and read any section directly, the way an app reads an API.
POST /mcp · search, get_sectionHow to connect
06
An AI that answers for the site
The Ask box answers visitors’ questions from my own writing, with links to the sources, and sticks to what the site covers.
Ask about my workSwitched on when Ask is live.
07
AI crawlers, invited
A robots.txt that names the AI crawlers and welcomes them, instead of leaving it to their defaults.
User-agent: GPTBot · Allow: /Open it

Connect an assistant

Any MCP client can read this site directly. Add a remote server with this address. It is read-only and needs no key.

https://lornefade.com/mcp

Try it

Enter any website. It checks seven signals that help AI read a site, and shows how it compares with this one.

Fade Digital builds every layer on this page for clients, and tracks whether ChatGPT, Gemini and Perplexity start citing them.