Semrush MCP Server

The Semrush MCP Server makes Semrush's databases available to AI assistants. You ask in plain language, for example about a domain's organic competitors, keywords with high search volume and low difficulty, or traffic trends across an industry, and the assistant pulls the matching reports straight from Semrush. MCP, the Model Context Protocol, is an open standard that assistants such as Claude, ChatGPT or Cursor use to call external tools. Semrush runs the server itself at https://mcp.semrush.com/v2/mcp. So you install nothing locally; you simply connect your client to that endpoint.

Which data does the server provide?

According to the developer documentation, three areas are available. The SEO API covers all endpoints for domain, keyword, backlink, subdomain, subfolder and URL reports, overview reports, competitors and data on ads and shopping ads. The Trends API delivers traffic and market data, depending on your subscription tier. From the Projects API, only read methods are exposed: the assistant can fetch your projects but cannot create or change them. That is a sensible boundary, because it means an assistant cannot alter settings in your account.

Where does the Semrush MCP Server work?

Semrush documents the setup for Claude on the web and desktop, Claude Code, ChatGPT, Cursor, VS Code, Google's Antigravity, Microsoft Copilot, Perplexity, Lovable and Notion. In Claude, ChatGPT and Perplexity, Semrush is available as a ready-made connector or app that you link with a few clicks. In developer tools such as Claude Code, Cursor or VS Code you enter the server address. According to the documentation, Semrush no longer supports the earlier route through the Gemini CLI and points to Antigravity instead.

Sign-in and security

OAuth 2.1 is the default: the client redirects you to the Semrush sign-in, you approve access, and you are done. There is no key to copy into a file. For clients without OAuth support, the server alternatively accepts an API key in a header. Such a key grants access to your account and should not end up in shared configuration files or repositories.

Which subscription do you need?

This deserves a closer look, because not every Semrush account has MCP access. According to the documentation, SEO data requires one of the Semrush One Starter, Semrush One Pro+, SEO Classic Pro or SEO Classic Guru subscriptions; these include an allowance of API units that renews with each billing cycle. If you are on SEO Classic Business or Semrush One Advanced, you additionally need an API unit package. Traffic and market data require a Trends API subscription. Every request uses units, depending on report type and data volume.

Cost

The server itself carries no extra charge, but queries run through your Semrush subscription and use API units. Exact terms and prices for subscriptions and unit packages are listed on Semrush's official pages.

Who is the Semrush MCP Server for?

It pays off for marketing and SEO teams that already use Semrush and want to get analyses done without detouring through exports: competitor analyses, keyword research for new content, market comparisons or quick answers to questions from management. Semrush itself notes that AI answers can be incomplete or wrong. So double-check important figures in the Semrush interface before you base decisions on them.

A Semrush subscription with MCP access (according to the documentation Semrush One Starter, Semrush One Pro+, SEO Classic Pro or SEO Classic Guru; with SEO Classic Business or Semrush One Advanced, an API unit package on top; for traffic data, a Trends API subscription) and an MCP client with HTTP transport.

Step 1: Check your subscription

Check which subscription your Semrush account has. According to Semrush, MCP access to SEO data comes with Semrush One Starter, Semrush One Pro+, SEO Classic Pro and SEO Classic Guru. With SEO Classic Business or Semrush One Advanced you additionally need an API unit package, and for traffic and market data a Trends API subscription.

Step 2: Add it to Claude Code

Open a terminal and run this command:

claude mcp add semrush https://mcp.semrush.com/v2/mcp -t http

Start Claude Code and enter:

/mcp

Select the Semrush server, click Authenticate and approve access in the browser.

Step 3: Alternatively in Claude on the web or ChatGPT

In Claude, open claude.ai, go to Customize and Connect your apps, search for Semrush, click Connect and confirm with Approve. In ChatGPT, you will find Semrush under Account, Settings, Apps; click Connect there, then Continue and Approve. In both cases you address the service in chat with @Semrush.

Step 4: Alternatively in Cursor

In Cursor, open Settings, Cursor Settings, Tools & MCP and add this entry:

{
  "mcpServers": {
    "semrush": {
      "url": "https://mcp.semrush.com/v2/mcp"
    }
  }
}

On first use, the Semrush sign-in opens in the browser.

Step 5: Alternatively in VS Code

Create the file .vscode/mcp.json in your project folder and add:

{
  "servers": {
    "semrush": {
      "url": "https://mcp.semrush.com/v2/mcp",
      "type": "http"
    }
  }
}

Step 6: Alternatively use an API key instead of OAuth

If your client does not support OAuth, send your Semrush API key in a header. In Claude Code it looks like this; replace YOUR_API_KEY with your key:

claude mcp add semrush https://mcp.semrush.com/v2/mcp -t http --header "Authorization: Apikey YOUR_API_KEY"

Treat the key like a password and do not store it in files that end up in a Git repository.

Step 7: Test the connection

Ask your assistant, for example:

Use Semrush and list the five most important organic competitors of example.com in the United States.

If a list of domains appears, the connection works. You can see your unit consumption in your Semrush account.

OAuth 2.1 by default with no extra configuration. Alternatively, a Semrush API key in the header "Authorization: Apikey ..." for clients without OAuth.

Read access to all SEO API endpoints (domain, keyword, backlinks, subdomain, subfolder, URL, overviews, competitors, ads), to the Trends API depending on subscription and to the read methods of the Projects API. Projects cannot be created or changed.

Requests go through mcp.semrush.com. Retrieved reports and project data enter the AI client context and therefore reach the model provider.

An assistant can trigger many queries without asking and use up API units. An API key in a header grants access to your account. According to Semrush, AI answers can be incomplete or wrong.