RankControl MCP Server

What is RankControl MCP Server?

The RankControl MCP server puts your RankControl AI SEO workspace inside Claude, Cursor, Codex or any MCP client, with 80+ tools for AI citations, content, rankings, analytics, link building and publishing. It is hosted at https://api.rctrl.com/mcp and uses browser sign-in, with no API keys for the remote connector.

RankControl is an AI SEO and content platform. It writes in-depth articles meant to be cited by ChatGPT, Perplexity, Claude, Gemini, Grok and Google AI Mode and to rank on Google and Bing, publishes them as native posts through your CMS (WordPress, Webflow, Shopify, Ghost, Notion, Wix, Framer, webhooks or a headless Content API), and tracks citations, rankings, backlinks and AI crawler activity per page.

How to use RankControl MCP Server?

You need a RankControl workspace. Only the workspace owner can approve the connection, and a workspace still in setup must finish onboarding first. RankControl's llms.txt describes the hosted endpoint as OAuth 2.1; on the consent screen you choose which areas the assistant may access.

Server URL:

https://api.rctrl.com/mcp

Claude Code

claude mcp add --transport http rankcontrol https://api.rctrl.com/mcp

Run /mcp and sign in to RankControl in your browser.

Claude Desktop

Open Settings, then Connectors, click Add custom connector, name it RankControl, paste https://api.rctrl.com/mcp, save and click Connect. Sign in and approve access in the browser. The same steps work on claude.ai.

Cursor

{
  "mcpServers": {
    "rankcontrol": {
      "url": "https://api.rctrl.com/mcp"
    }
  }
}

Add this to ~/.cursor/mcp.json and complete the browser sign-in.

VS Code

{
  "servers": {
    "rankcontrol": {
      "type": "http",
      "url": "https://api.rctrl.com/mcp"
    }
  }
}

Add this to .vscode/mcp.json and start the server.

Local server (optional)

RankControl also ships the CLI and MCP server in one npm package. Run rankcontrol login once, then:

claude mcp add rankcontrol -- npx rankcontrol mcp

Other clients can run npx rankcontrol mcp with an API key in RANKCONTROL_API_KEY; the local server acts under that key's scopes.

Key features of RankControl MCP Server

• 80+ tools grouped by area: Visibility, Content, Reports, Analytics, Insights, Repurposing, Link building, Social, Brand memory, Team and Support.
• Consent-screen access categories, from reading AI citations, articles and analytics to planning content, managing outreach and social drafts, publishing approved articles and managing settings.
• Dry run by default: writing, publishing, social push, outreach sends and team invites return a preview until called with confirm: true.
• Removals, archiving, hiding and team changes wait for your approval inside RankControl.
• A 2-hour hold: after the agent reads external text (prospect replies, AI answers, social threads, competitor pages), every change waits for approval.
• Rate caps of 5 to 10 calls per minute on costly tools.
• contact_support files a report to the RankControl team from the assistant.

Use cases of RankControl MCP Server

• Visibility check: "How often did ChatGPT and Perplexity cite us this week, and which competitors were cited instead?"
• Content planning: "Plan next month's content calendar around the queries where we are not cited yet."
• Drafting and publishing: "Draft an article on our top opportunity query and show me the preview before publishing to WordPress."
• Analytics: "Which pages got AI referral traffic from Claude and Gemini last month?"
• Repurposing: "Turn yesterday's article into LinkedIn and X drafts."

Troubleshooting RankControl MCP Server

• You cannot approve the connection. Only the workspace owner can approve, and onboarding must be finished first.
• Nothing changes after a write. Writes are dry runs by default; the tool must be called again with confirm: true.
• Changes are stuck waiting for approval. After the agent reads external text, all changes wait for approval for 2 hours; approve them in RankControl.
• Calls are throttled. Costly tools are capped at 5 to 10 calls per minute. Space out requests.
• Still stuck. Use /contact on rctrl.com or ask the agent to call contact_support.