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

# What is the IQRush MCP connector?

> A hosted Model Context Protocol server that lets ChatGPT, Claude, and other AI assistants read your IQRush measurement data.

The IQRush MCP connector is a hosted [Model Context Protocol](https://modelcontextprotocol.io) server. It gives an AI assistant read access to your IQRush projects, runs, and measurements. Instead of opening the dashboard, you ask a question in the assistant you already use, and the assistant calls IQRush in the background.

## How it works

1. You ask a question, such as "What is our citation share on the latest run?"
2. Your assistant (Claude, ChatGPT, or another MCP client) calls the IQRush server.
3. IQRush returns measurements for a project and run you can access.
4. The assistant explains the result in plain language.

Ask the assistant to keep the confidence intervals, stability checks, and change assessments that IQRush returns. They show how strong the evidence is. See [decision-grade results](/developers/guides/10-stability).

## Server URL

The production remote URL is:

```text theme={null}
https://api.prod.gso.iqrush.ai/api/mcp/mcp
```

You can also copy it from **Settings → MCP → Connect** in IQRush. A Dev environment uses a different URL and different data, so use the URL from the environment that holds your projects.

## Choose your assistant

| Assistant | How it connects | Guide |
| - | - | - |
| Claude Desktop | One-line installer, then a browser sign-in link | [Connect Claude Desktop](/mcp/connect-claude-desktop) |
| Claude.ai | Add a custom connector with the server URL | [Connect Claude.ai or Claude Code](/mcp/connect-claude-ai) |
| Claude Code | `claude mcp add` with the server URL | [Connect Claude.ai or Claude Code](/mcp/connect-claude-ai) |
| ChatGPT | Create a custom app with the server URL | [Connect ChatGPT](/mcp/connect-chatgpt) |
| Other MCP clients | Add the server URL in the client's MCP settings | [Build a custom MCP client](/developers/guides/05-custom-mcp-client) |

## What you can ask

* Compare your brand's citation share with competitors across AI engines.
* See how your mention rate changed between runs.
* Find sources that cite your competitors but not you.
* Review how AI answers describe your brand, including sentiment.
* Check whether a change between two runs is statistically meaningful.
* Combine AI visibility with Google Analytics 4 traffic, if you have connected a property.

For ready-made prompts, see [Questions and reports to explore](/guides/31-questions-and-reports).

## What the connector can and cannot do

The connector **reads** your projects, runs, measurements, brand context, and connected Google data. It exposes 43 tools, listed in the [tool index](/developers/guides/23-tool-index).

It does **not** create or change projects, runs, settings, or brand context, and it does not publish content. It only returns data for projects the signed-in user can already access.

## Before you connect

* An IQRush account.
* At least one project with a completed run.
* For Claude Desktop: Python 3.11 or later, or `uv`.
* For Google Analytics tools: a Google Analytics 4 property connected in IQRush **Settings → Integrations**.

## Next steps

* [Connect your assistant](/mcp/connect-claude-desktop), then [make your first MCP calls](/developers/guides/02-mcp-quickstart).
* [How sign-in and permissions work](/developers/guides/04-authentication).
* [Browse the tool reference](/developers/guides/23-tool-index).


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