> ## Documentation Index
> Fetch the complete documentation index at: https://areyoufoundbyai.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect Are you found by AI? to Claude, Cursor, ChatGPT and Gemini

> Set up Are you found by AI? in the Claude app, Claude Code and Cursor over MCP, and in Claude Projects, ChatGPT, Perplexity and Gemini with a standing instruction.

The Are you found by AI? MCP integration gives your AI assistant a live view of your visibility measurements. Instead of copying scores out of your console and pasting them into a chat, your assistant queries the data directly, mid-conversation, whenever the question comes up, and always answers from this week's numbers. This guide covers getting your endpoint and connecting it to the main AI tools.

## Get your MCP endpoint

MCP access is included on the free plan as well as Pro. Each monitored site has its own token, and your console's **Connect your AI** section has the 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 monitoring email link: it is the part after `/monitor/`.

The recommended form is the static endpoint with the token in a header:

```bash theme={null}
curl -X POST https://areyoufoundbyai.com/mcp \
  -H "authorization: Bearer YOUR-TOKEN" \
  -H "content-type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
```

`x-api-key: YOUR-TOKEN` works too. URLs end up in server logs, proxies and browser history, and headers do not. For clients that only take a URL, the path form keeps working:

```text theme={null}
https://areyoufoundbyai.com/mcp/YOUR-TOKEN
```

The token is your credential. Treat it like a password: share it with your own AI tools and teammates you trust, but never post it in a public repository or shared document.

<Note>
  **No monitor yet?** Try the demo endpoint, connected to our own live monitoring of areyoufoundbyai.com: `https://areyoufoundbyai.com/mcp/demo`. It carries all 20 tools, needs no key and no signup, and its `request_rescan` returns a worked example and queues nothing. Then run your own free scan at [areyoufoundbyai.com](https://areyoufoundbyai.com) and your monitor gets an endpoint of its own.
</Note>

## Connect to Claude Code

Open a terminal and run:

```bash theme={null}
claude mcp add --transport http found-by-ai https://areyoufoundbyai.com/mcp/YOUR-TOKEN
```

Replace `YOUR-TOKEN` with your token, or use `demo` to try the public demo first:

```bash theme={null}
claude mcp add --transport http found-by-ai https://areyoufoundbyai.com/mcp/demo
```

Any Claude Code session can then query your measurements directly. If a coding agent builds your site, it can read your fix plan over this connection, apply the fixes in your repository with your approval, and re-measure.

## Connect to the Claude app

In the Claude app, open **Settings**, then **Connectors**, then **Add custom connector**, and paste your endpoint URL. Any chat can then ask "what moved in my AI visibility this week?" and Claude queries it live.

## Connect to Claude Projects

Claude Projects support persistent instructions that apply to every new conversation inside the project, so your assistant starts each chat with your current data and you paste nothing.

Create or open a Project in Claude, then find its custom instructions or project knowledge setting (the exact label changes as Claude updates). Paste the following standing instruction, replacing `YOUR-TOKEN` with your token:

```text theme={null}
At the start of every new conversation in this project, fetch https://areyoufoundbyai.com/monitor/YOUR-TOKEN.json (or the plain-text version at https://areyoufoundbyai.com/monitor/YOUR-TOKEN/context.md) and use it as my current Are you found by AI? data, my AI-visibility score, tracked buyer questions, priority fixes and citation targets, before answering anything about my AI visibility, competitors, or what to do next. Re-fetch it fresh every new chat; don't rely on a cached copy from an earlier conversation.
```

The standing instruction makes your AI read your data at the start of a chat. An MCP connection through the Claude app lets it query your measurements mid-conversation as well.

## Connect to ChatGPT Projects

ChatGPT Projects support instructions that apply to every new chat within the project. Open a Project in ChatGPT and paste the same standing instruction into the project's instructions field.

<Note>
  If you are building a custom GPT and want it to query your data on demand mid-conversation, rather than only at the start, that needs an Action wired to the JSON endpoint, which is more setup. Email [hello@techhorizonlabs.com](mailto:hello@techhorizonlabs.com) if you want a hand.
</Note>

## Connect to Perplexity Spaces

Perplexity Spaces support custom instructions the same way. Create a Space dedicated to your business and paste the standing instruction into its custom instructions. Every new conversation in that Space starts by fetching your latest measurements.

## Connect to Gemini Gems

Gemini Gems support a persistent instructions field. Create a Gem for your business and paste the standing instruction into it. Look for **Instructions**, **Custom instructions** or **Project knowledge**, depending on the current Gemini interface.

## Connect to Cursor

In Cursor, add a custom MCP server with HTTP transport and use your endpoint:

```text theme={null}
https://areyoufoundbyai.com/mcp/YOUR-TOKEN
```

Or use the one-click **Add to Cursor** link in your console's **Connect your AI** section, which fills in the configuration for you.

## What you can ask

Once your assistant is connected, ask it anything you would ask a colleague who had your console open. With a live connection these questions work mid-conversation, weeks after setup, and always answer from current data:

<Accordion title="Score and trend questions">
  * "What moved in my AI visibility this week, and what is the single most useful thing I can do about it?"
  * "Show me my readiness subscores and compare them to last week."
  * "What is my off-site Footprint score, and what sits behind it?"
</Accordion>

<Accordion title="Competitor and market questions">
  * "Which competitors are being named instead of me?"
  * "What is my share of voice against the rivals the engines name?"
  * "Which buyer questions am I still invisible on, and who is AI naming instead of me on those?"
</Accordion>

<Accordion title="Fix plan and action questions">
  * "What should I fix first?"
  * "Read my fix plan and sort the items into ones I can apply in my repository and ones I need to handle elsewhere."
  * "Which sources do the engines actually cite in my category, and which ones do I appear on?"
</Accordion>

<Accordion title="Content and briefing questions">
  * "Draft a short update for my business partner: our current scores, the trend, and any new mentions this month."
  * "Build me a post brief from this week's verified data: the questions I'm absent from and the sources the engines read."
  * "What new independent mentions of my business have been found recently?"
</Accordion>

## Full list of available tools

The demo endpoint lists 20 tools: 19 read tools and one action, `request_rescan`, which queues a fresh measurement within your plan's cap and changes nothing else. On the demo, `request_rescan` returns a worked example without queuing anything. Your own token lists more: three saved-only reads (`get_balance`, `get_weekly_brief` and `get_grants`), the workspace reads such as `get_token_balance` for your shared token balance, and, when the account owner has turned on editing, workspace actions such as `quote_token_run` and `confirm_token_run`.

See the [MCP Tools Reference](/docs/mcp/tools) for every tool and its parameters.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.