> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryxlr8.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect the MCP server

> Use XLR8.ai from Claude, ChatGPT, Cursor, Claude Code, VS Code, or Codex. Add the server URL, sign in with your XLR8.ai account, and start asking.

The XLR8.ai MCP server lets your AI assistant work in your XLR8.ai account. Ask it to build a query set, start an experiment, break down which sources AI models cite, draft on-brand content, or find Reddit threads to join, all from the chat you already use.

To connect, add the server URL to your assistant and sign in with the same account you use at [app.tryxlr8.ai](https://app.tryxlr8.ai). There is nothing to install.

```text theme={null}
https://mcp.tryxlr8.ai/mcp
```

> [card] [Claude](#claude)
>
> Web, desktop, and Cowork

> [card] [ChatGPT and Codex](#chatgpt-desktop-and-codex)
>
> ChatGPT desktop app, Codex CLI, and IDE extension

> [card] [Cursor](#cursor)
>
> One-click install

> [card] [Claude Code](#claude-code)
>
> One terminal command

> [card] [VS Code](#vs-code)
>
> With GitHub Copilot

> [card] [Other clients](#other-clients)
>
> Any client that supports remote MCP

## Set up your client

### Claude

Add XLR8.ai as a custom connector. It works in Claude on the web, Claude Desktop, and Cowork. For more detail, see [Claude's custom connector guide](https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp).

> [tab] Free, Pro, and Max
>
> **1. Add a custom connector**
>
> Go to **Customize → Connectors**, select **+ Add**, then select **Add custom connector**.
>
> **2. Enter the server details**
>
> Name the connector `XLR8.ai`, paste `https://mcp.tryxlr8.ai/mcp`, and select **Continue**.
>
> **3. Sign in**
>
> Keep the authentication settings Claude detects, choose **Sign in now**, and select **Add**. Sign in with your XLR8.ai account.
>
> **4. Turn it on in a chat**
>
> In a conversation, select **+**, open **Connectors**, and turn on XLR8.ai.
>
> Free plans can add one custom connector.

> [tab] Team and Enterprise
>
> An owner adds the connector once for the organization. Each member then connects their own XLR8.ai account.
>
> **1. Owner: add the connector**
>
> Go to **Organization settings → Connectors**, select **Add**, hover over **Custom**, and select **Web**. Name it `XLR8.ai`, paste `https://mcp.tryxlr8.ai/mcp`, keep the detected authentication settings, and select **Add**.
>
> **2. Member: connect your account**
>
> Go to **Customize → Connectors**, find XLR8.ai, and select **Connect**. Sign in with your XLR8.ai account.
>
> **3. Turn it on in a chat**
>
> In a conversation, select **+**, open **Connectors**, and turn on XLR8.ai.

### ChatGPT desktop and Codex

The ChatGPT desktop app, the Codex CLI, and the Codex IDE extension share one MCP configuration. Add XLR8.ai in one of them and it is available in all three. For more detail, see the [Codex MCP documentation](https://developers.openai.com/codex/mcp/).

> [tab] ChatGPT desktop app
>
> **1. Add the server**
>
> Open **Settings → MCP servers** and select **Add server**. Name it `xlr8`, choose **Streamable HTTP**, and paste `https://mcp.tryxlr8.ai/mcp`.
>
> **2. Restart and sign in**
>
> Save the server and select **Restart**. Then select **Authenticate** next to `xlr8` and sign in with your XLR8.ai account.

> [tab] Codex CLI
>
> **1. Add the server**
>
> ```bash theme={null}
> codex mcp add xlr8 --url https://mcp.tryxlr8.ai/mcp
> ```
>
> **2. Sign in**
>
> ```bash theme={null}
> codex mcp login xlr8
> ```
>
> Your browser opens so you can sign in with your XLR8.ai account.

> [tab] Codex IDE extension
>
> **1. Add the server**
>
> Open the gear menu, select **MCP servers**, then select **Add server**. Name it `xlr8`, choose **Streamable HTTP**, and paste `https://mcp.tryxlr8.ai/mcp`.
>
> **2. Restart and sign in**
>
> Save the server and select **Restart extension**. Then select **Authenticate** and sign in with your XLR8.ai account.

### Cursor

For more detail, see the [Cursor MCP documentation](https://cursor.com/docs/mcp).

**1. Add the server**

[Add XLR8.ai to Cursor](cursor://anysphere.cursor-deeplink/mcp/install?name=xlr8\&config=eyJ1cmwiOiJodHRwczovL21jcC50cnl4bHI4LmFpL21jcCJ9) and confirm the install when Cursor asks.

To add it by hand, put this in `~/.cursor/mcp.json` to use it in every project, or in `.cursor/mcp.json` for one project:

```json theme={null}
{
  "mcpServers": {
    "xlr8": {
      "url": "https://mcp.tryxlr8.ai/mcp"
    }
  }
}
```

**2. Sign in**

Open **Customize** in Cursor's sidebar, turn on XLR8.ai, and sign in with your XLR8.ai account when your browser opens.

### Claude Code

For more detail, see the [Claude Code MCP documentation](https://docs.anthropic.com/en/docs/claude-code/mcp).

**1. Add the server**

```bash theme={null}
claude mcp add --transport http xlr8 https://mcp.tryxlr8.ai/mcp
```

To use XLR8.ai in every project, not only the current one, add `--scope user` after `xlr8`.

**2. Sign in**

Start Claude Code, run `/mcp`, select `xlr8`, and sign in with your XLR8.ai account in the browser.

### VS Code

Use XLR8.ai with GitHub Copilot in VS Code. For more detail, see the [VS Code MCP documentation](https://code.visualstudio.com/docs/agent-customization/mcp-servers).

**1. Add the server**

Open the Command Palette (`Cmd+Shift+P` on Mac, `Ctrl+Shift+P` on Windows and Linux) and run **MCP: Add Server**. Choose **HTTP**, paste `https://mcp.tryxlr8.ai/mcp`, and enter `xlr8` as the server ID.

**2. Choose where to save it**

Select **Global**, then **Copilot Global**, to use XLR8.ai in every workspace.

**3. Sign in**

When VS Code asks, allow the sign-in and sign in with your XLR8.ai account.

### Other clients

Any client that supports remote MCP servers can connect. Use these settings:

* **URL:** `https://mcp.tryxlr8.ai/mcp`
* **Transport:** Streamable HTTP (some clients call it HTTP)
* **Authentication:** OAuth. Sign in with your XLR8.ai account when the client asks.

If your client takes a JSON configuration:

```json theme={null}
{
  "mcpServers": {
    "xlr8": {
      "url": "https://mcp.tryxlr8.ai/mcp"
    }
  }
}
```

If your client can only run local servers, connect through [mcp-remote](https://www.npmjs.com/package/mcp-remote). It needs Node.js and opens your browser so you can sign in.

```json theme={null}
{
  "mcpServers": {
    "xlr8": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.tryxlr8.ai/mcp"]
    }
  }
}
```

## Check the connection

Start a new chat and ask:

Test the connection

List my XLR8.ai organizations.

The assistant should list each organization you belong to and your role in it. If you belong to more than one, tell it which one to use, for example "Use Acme for this chat."

## What you can ask

Copy one of these prompts and replace the text in brackets.

**Start a visibility experiment**

Create a visibility query set for \[company] at \[domain] with about 20 queries. Show me the categories and queries first. After I approve them, save the query set and start an experiment with it.

**See which sources AI models cite**

For the experiment \[experiment name], group the cited sources by category and tell me which third-party sites appear most often. Suggest where we should work on getting mentioned.

**Draft on-brand content**

Write a blog post about \[topic] for \[organization]. Follow our brand guidelines and add internal links to our site.

**Check a draft for AI detection**

Scan each section of \[content title] with GPTZero. For any section that reads as AI-written, show me a humanized version and wait for my approval before you save it.

**Find Reddit threads to join**

Find recent Reddit threads where people ask about \[topic]. Pick the three most relevant and draft a helpful reply for each.

> [!TIP]
>
> Some requests start work in your account, such as running an experiment, generating content, or sending an outreach email. Keep tool approval on in your assistant so you can review each action before it runs.

## Connect with an API key

Use an API key when your client can't open a browser to sign in, for example an agent that runs on a server. Contact XLR8.ai to get an API key for your account.

The client sends the key as a bearer token on every request:

```http theme={null}
Authorization: Bearer <your_api_key>
```

```json Cursor theme={null}
{
  "mcpServers": {
    "xlr8": {
      "url": "https://mcp.tryxlr8.ai/mcp",
      "headers": {
        "Authorization": "Bearer ${env:XLR8_API_KEY}"
      }
    }
  }
}
```

```bash Claude Code theme={null}
claude mcp add --transport http xlr8 https://mcp.tryxlr8.ai/mcp \
  --header "Authorization: Bearer $XLR8_API_KEY"
```

```toml Codex theme={null}
[mcp_servers.xlr8]
url = "https://mcp.tryxlr8.ai/mcp"
bearer_token_env_var = "XLR8_API_KEY"
```

```json Other clients theme={null}
{
  "mcpServers": {
    "xlr8": {
      "url": "https://mcp.tryxlr8.ai/mcp",
      "headers": {
        "Authorization": "Bearer <your_api_key>"
      }
    }
  }
}
```

The Cursor, Claude Code, and Codex examples read the key from an `XLR8_API_KEY` environment variable, so the key stays out of your config files. Set that variable before you start the client. If your client doesn't fill in environment variables, paste the key in place of the variable.

In Claude, add the custom connector as described above, choose **No sign in**, and add a request header named `Authorization` with the value `Bearer <your_api_key>`.

> [!WARNING]
>
> On Claude Team and Enterprise plans, everyone who uses a connector set up with an API key acts as the key's owner. Use sign-in instead so each member connects their own account.

If your key isn't tied to one organization, tell the assistant which organization to use. The [System Reporting API](https://docs.tryxlr8.ai/api-reference/overview.md) key is a different key and doesn't work with the MCP server.

## Troubleshooting

> [accordion] Sign-in fails or never finishes
>
> Remove XLR8.ai from your client, add it again, and sign in with the same account you use at [app.tryxlr8.ai](https://app.tryxlr8.ai).
>
> If you connect through `mcp-remote`, also delete the `.mcp-auth` folder in your home directory to clear its saved sign-ins, then try again.

> [accordion] The assistant says it isn't authorized
>
> Your saved sign-in expired or is no longer valid. Disconnect XLR8.ai in your client's connector or MCP settings and connect again.
>
> With an API key, check that the header contains the key itself and not the name of the environment variable. Copy the key again with no extra spaces.

> [accordion] XLR8.ai doesn't show up in the chat
>
> Make sure the server is turned on. In Claude, select **+** in the conversation and check **Connectors**. In Cursor, check **Customize**. If you edited a config file, quit the client completely and open it again.

> [accordion] My organization is missing
>
> The assistant only sees organizations your account belongs to. Ask an admin to [invite you](https://docs.tryxlr8.ai/account/authentication.md#inviting-team-members), or sign in with the account that already has access.

> [accordion] My company blocks the connection
>
> Your IT team may restrict external MCP servers. Ask them to allow `mcp.tryxlr8.ai`. On Claude Team and Enterprise plans, only owners can add custom connectors.

## FAQ

> [accordion] Why does the assistant ask to run search\_tools and call\_tool?
>
> XLR8.ai has a large set of tools. The assistant starts with a few, uses `search_tools` to find the one your request needs, and runs it with `call_tool`. Expand the approval request to see which XLR8.ai tool it will run and what it will send.

> [accordion] What can the assistant access?
>
> Only organizations your XLR8.ai account belongs to. In those organizations, it can work with query sets, experiments, citation results, content, brand guidelines, Reddit listening, outreach, and videos.

> [accordion] How do I disconnect?
>
> Remove XLR8.ai from your client's connector or MCP settings. In Claude, go to **Customize → Connectors** and select **Remove**.

## Related

> [card] [Experiments](https://docs.tryxlr8.ai/experiments/overview.md)
>
> Track how AI models talk about your brand.

> [card] [Query generation](https://docs.tryxlr8.ai/query-generation.md)
>
> Build the query sets your experiments run.

> [card] [Content generation](https://docs.tryxlr8.ai/content/generation.md)
>
> Create content designed to improve your AI visibility.

> [card] [Social listening](https://docs.tryxlr8.ai/intelligence/social-listening.md)
>
> Find Reddit conversations worth joining.


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