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

# Connect with MCP

> Add the Laurence MCP server to Claude, ChatGPT, Grok, Cursor, VS Code or another agent, sign in, and start asking about your account

The Laurence MCP server gives your AI agent a live, read-only connection to your company's Laurence data. MCP (Model Context Protocol) is the open standard AI agents use to call outside tools. Setup takes about a minute. The quickest way is the platform's [Settings → Agents](https://www.laurence.com/platform/settings/agents) page: pick your agent, click its button, and it opens the app ready to add Laurence.

## Before you start

You need a Laurence login with access to at least one company, and an agent that supports remote MCP servers.

## Set it up

**1. Copy the server URL**

```text theme={null}
https://mcp.laurence.com/mcp
```

**2. Add it to your agent**

Follow the steps for your agent in the next section.

**3. Sign in and allow**

Your agent opens a Laurence sign-in page. Sign in with your Laurence login, choose your company and click Allow.

## Your agent

### Claude

On claude.ai, open [Settings → Connectors](https://claude.ai/settings/connectors), choose **Add custom connector**, paste the server URL and click **Add**, then **Connect**. The connector also works in Claude Desktop and the mobile apps.

### ChatGPT

In ChatGPT, open **Settings → Apps & Connectors**. On a paid plan with developer mode on (under **Advanced**), create a connector, paste the server URL and choose OAuth.

### Grok

On grok.com, open [Connectors](https://grok.com/connectors), click **New Connector**, choose **Custom** and paste the server URL. In Grok Bot you can also say "Add this MCP server" followed by the URL.

### Cursor

Open the [Settings → Agents](https://www.laurence.com/platform/settings/agents) page and click **Add to Cursor**. Cursor opens and asks to install Laurence; then click **Connect** next to it in **Cursor Settings → Tools & MCP**. To add it by hand instead, put this in your MCP settings:

```json theme={null}
{
  "mcpServers": {
    "laurence": {
      "url": "https://mcp.laurence.com/mcp"
    }
  }
}
```

### VS Code

Click **Add to VS Code** on the [Settings → Agents](https://www.laurence.com/platform/settings/agents) page. VS Code asks to install Laurence; start the server when it asks, to sign in.

### Claude Code

Run:

```bash theme={null}
claude mcp add --transport http laurence https://mcp.laurence.com/mcp
```

Then run `/mcp` in Claude Code to sign in.

### Codex

Run:

```bash theme={null}
codex mcp add laurence --url https://mcp.laurence.com/mcp
```

Then sign in with:

```bash theme={null}
codex mcp login laurence --scopes laurence:mcp
```

### Le Chat, Perplexity and Copilot Studio

Add a custom MCP connector with the server URL and choose OAuth: in Le Chat under **Connectors**, in Perplexity under **Settings → Connectors** (paid plans), and in Copilot Studio under your agent's **Tools → Add a tool → Model Context Protocol**, with OAuth 2.0 dynamic discovery.

### Agent builders and API scripts

Most hosted agent builders (OpenAI Agent Builder, Zapier, Make, n8n Cloud) open the same sign-in page when you add the server URL. Tools that cannot open a sign-in page, such as self-hosted n8n, scripts on the Claude, xAI or OpenAI APIs, and Manus, use a personal access token instead:

1. On [Settings → Agents](https://www.laurence.com/platform/settings/agents), under **Tokens for agent builders**, click **Create token**, name it and choose the company it reads.
2. Copy the token (it starts with `lmcp_`). It is shown once.
3. Give your tool the server URL and send the token as a Bearer token in the `Authorization` header.

You can see when each token was last used and revoke it at any time.

## What your agent can answer

Your agent sees the same read-only tools Ask Laurence uses on the platform, grouped here by the questions they answer:

| Area | What your agent can look up |
| - | - |
| Ad performance | Ad spend, ad sales, orders, ROAS and CPC by day or month; branded and non-branded splits; performance by placement; hour-by-hour traffic for a day |
| Search terms | What shoppers searched before clicking your ads, with what each search cost and sold |
| Sales and products | Total sales across all channels, daily revenue, Best Sellers Rank history, product details, FBA inventory, and the deals and coupons you ran |
| Campaigns and targeting | Your campaigns, ad groups, keywords and negative keywords, which campaigns advertise a product, and how a product's keywords are grouped |
| Market | Search volume and your share of impressions, clicks and purchases on a search, and how popular a search is |
| Profit and customers | Product margins, net profit, customer acquisition cost and long-term customer value |
| What Laurence did | Keywords added and paused, bid changes, each target's history, your account's targets and the KPI scorecard you see on your dashboard |
| Your account | Your shared Slack channel with Laurence and the notes from your calls with us |
| The platform | Where things are on the Laurence platform and what each number on a page means |

If your login covers more than one company, ask your agent which companies it can see, or name the company in your question.

## Try asking

* "How much did we spend on ads last month, and what did it return?"
* "Which search terms cost the most last week without a sale?"
* "How did branded and non-branded ads compare over the last 30 days?"
* "What did Laurence change on our account this week?"
* "Which products are running low on FBA inventory?"

## Access and security

* **Read only.** No tool can change a campaign, bid, budget or listing.
* **One company per connection.** The company you choose at sign-in, or when you create a token, is the only one the connection can see. Connect again to choose a different one.
* **Checked on every call.** Each request is checked against your current Laurence access, so if you are removed from a company, the connection stops answering for it.
* **Disconnect any time** by removing the Laurence server in your agent's settings, or by revoking its token on the Settings → Agents page.


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