Skip to main content
The Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools mid-conversation, pulling fresh data on demand rather than relying on whatever you paste into the chat. The Are you found by AI? MCP server exposes your measurements (scores, saved engine answers, fix plan, competitors, citation sources) as tools: the public demo lists 20 (19 read tools and one action), and a customer token lists more, including the shared token balance and saved client reports. Connect it once and every conversation starts from your latest numbers, not a stale snapshot.

​
What you can do with the MCP integration

Once connected, you can ask your AI assistant about your AI visibility in plain language, without opening a dashboard or copying a report. The integration lets you:
  • Re-measure: queue a fresh measurement from inside a conversation with request_rescan (capped by your plan), then read the new scores a few minutes later. On a token account, you confirm the token price in your workspace first
  • Check your scores: ask for your current AI Visibility and AI Readiness scores, the subscores, and the separate off-site Footprint score
  • Get the prioritised fix list: ask what to fix first and receive the fixes from your latest deep scan, highest impact first
  • Read verbatim engine answers: see the saved answers ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek or Google AI Overviews gave when a buyer asked about your category, with the searches the engine ran first
  • Track competitors: find out which businesses AI names instead of you on the questions you are missing
  • Surface citation sources: discover which third-party domains the engines cite in your market
Because the tools query your saved measurements, the answers reflect your latest scan, not a model’s general knowledge about your business.

​
Your MCP endpoint

Each monitored site has its own token. The recommended form is the static endpoint with the token in a header:
x-api-key: YOUR-TOKEN works as well. A header is the better habit: URLs end up in server logs, proxies and browser history, and headers do not. The path form keeps working for any client that only takes a URL, or anything already pointed at it:
MCP access is included on the free plan as well as Pro.
Try it without an account. The demo endpoint is connected to our own live monitoring of areyoufoundbyai.com, real production data, no key and no signup:
Ask it “how visible is this business to AI, and what would you fix first?” to see what your own connected assistant would read.

​
Find your token

Your console’s Connect your AI section has your endpoint and setup steps pre-filled, with one-tap copy buttons. Your token is also part of any private link you already have, such as your console link or a link in a monitoring email: it is the part after /monitor/, as in areyoufoundbyai.com/monitor/YOUR-TOKEN. No monitor yet? Run the free scan at areyoufoundbyai.com and your monitor gets an endpoint of its own. Keep the token private. It is effectively a password to your own data: anyone who holds it can read your measurements. Never commit it to a public repository or paste it into a shared document. Sharing it with your own AI tool, or a teammate you trust, is what it is for.

​
Quick connect for Claude Code

The fastest way to try it is one terminal command in Claude Code, pointed at the demo:
For your own measurements, swap demo for your token. Then ask:
Claude Code will call the MCP server and answer from the live measurements rather than guessing. In the Claude app, add the endpoint under Settings, then Connectors, then Add custom connector. See the Connect guide for every client.

​
Why connect instead of copy-pasting

Pasting a weekly report into a chat works once, but every new conversation starts from zero, with no memory of what you already discussed and no guarantee the numbers are current. The MCP connection fixes three things at once:
  • No copy-pasting: your assistant queries the measurements whenever it needs a number
  • Always current: the same endpoint queried today answers from today’s data; queried in six weeks it answers from the data then
  • Fixes applied in context: a coding agent such as Claude Code can read your fix plan over this connection, apply the fixes in your repository with your approval, and re-measure

Connect Guide

Step-by-step setup for the Claude app, Claude Code, Cursor and other MCP clients, plus standing instructions for Claude Projects, ChatGPT, Perplexity and Gemini.

Available Tools

Full reference for every MCP tool and its parameters, plus the agency network catalogue.