RankReactor

RankReactor MCP

Config-only listing package for RankReactor's hosted Model Context Protocol server. This repository does not contain application or backend code.

RankReactor is a set-and-forget SEO product. You add a live website as a project. RankReactor reads the site, then can generate SEO articles and (on higher plans) UGC-style videos, and it tracks keywords, analytics, technical SEO, and a site audit.

MCP URL (use this exact host): https://www.rankreactor.ai/mcp

Always use www. The bare domain redirects and is not the MCP endpoint.

Plans: Free (1 lifetime article per project), Starter $39/mo, Pro $79/mo, Ultra $159/mo. UGC video is available on Pro and Ultra.

The hosted tools let an assistant read project status and stored SEO metrics, list generated articles and videos, refresh metrics, and queue new articles or UGC videos (write actions ask for confirmation). The current tool list is documented at Use RankReactor with your AI assistant.

Connect

Do this before adding the MCP URL:

1. Create a RankReactor account if you need one. A Free plan is available.
2. Create or select the project the assistant should use.
3. Open API Key in the dashboard sidebar and choose Generate. Copy the key now; RankReactor shows the full value once.

Then add https://www.rankreactor.ai/mcp in your client. An OAuth-capable client opens Connect RankReactorAI in the browser. Paste the project key and choose Allow. That window is not a Google, magic-link, or other RankReactor account sign-in.

OAuth keeps the project key out of client config (the files in this repo only contain the public URL). It does not remove the key-generation step. Regenerating or revoking the key in the dashboard stops existing connections until you paste the new key.

Clients that do not support MCP OAuth can use the same hosted Streamable HTTP URL with Authorization: Bearer <your-project-key>. Use that placeholder or the client's own secret prompt; do not commit a live key. JSON field names vary by client (url, httpUrl, or type: http). This repository does not ship a local stdio server.

Claude (claude.ai custom connector)

Remote MCP custom connectors work on Claude, Cowork, and Claude Desktop (Anthropic help). Free Claude accounts can add one custom connector.

One-click: Add RankReactor in Claude

Pro / Max

1. Open Customize → Connectors.
2. Click +, then Add custom connector.
3. URL: https://www.rankreactor.ai/mcp
4. Leave Advanced settings empty unless Claude asks for a client id.
5. Click Add. When Connect RankReactorAI opens, paste the project key and choose Allow.

Team / Enterprise: an Owner adds the URL under Organization settings → Connectors (Add → Custom → Web). Members then open Customize → Connectors and click Connect. When Connect RankReactorAI opens, paste the project key and choose Allow.

Enable the connector per conversation from + → Connectors.

Claude Code

From Claude Code MCP docs:

claude mcp add --transport http rankreactor https://www.rankreactor.ai/mcp

That writes a local-scope entry. Use --scope user for every project, or --scope project to write .mcp.json in the repo.

Then run claude, type /mcp, select rankreactor, and Authenticate. When Connect RankReactorAI opens, paste the project key and choose Allow.

Grok (grok.com)

From xAI Connectors:

1. Open grok.com/connectors.
2. Click New Connector, then Custom.
3. Enter https://www.rankreactor.ai/mcp. When Connect RankReactorAI opens, paste the project key and choose Allow.

On Grok Business / Enterprise, a team admin may need to provision the connector first in console.x.ai (Grok Business → Connectors → Add Connector → Other).

A Grok Build marketplace listing is a separate follow-up (see the pull request for the catalog entry text). You can still point Grok Build at this repo once it is listed, or add the same URL as a custom connector.

Cursor

Install link (opens Cursor; MCP install links):

Add RankReactor in Cursor

Deeplink equivalent:

cursor://anysphere.cursor-deeplink/mcp/install?name=rankreactor&config=eyJ1cmwiOiJodHRwczovL3d3dy5yYW5rcmVhY3Rvci5haS9tY3AifQ==

Manual mcp.json (Cursor MCP docs): project file .cursor/mcp.json, or ~/.cursor/mcp.json for every project.

{  "mcpServers": {    "rankreactor": {      "url": "https://www.rankreactor.ai/mcp"    }  }}

On first use, Cursor opens Connect RankReactorAI. Paste the project key and choose Allow. If OAuth does not open, the same hosted URL can take headers.Authorization as Bearer <your-project-key> (do not commit a live key).

This repo also ships a Cursor plugin manifest at .cursor-plugin/plugin.json plus mcp.json for Cursor Marketplace review, and a root .mcp.json for cursor.directory auto-detect.

Gemini CLI

From Gemini CLI extensions and the extension reference:

gemini extensions install https://github.com/rankreactorai/rankreactorai-mcp

The root gemini-extension.json registers the remote server with httpUrl. After install, start a new CLI session. If authentication is needed, run /mcp auth rankreactor. When Connect RankReactorAI opens, paste the project key and choose Allow.

You can also add the same server in ~/.gemini/settings.json without the extension:

{  "mcpServers": {    "rankreactor": {      "httpUrl": "https://www.rankreactor.ai/mcp"    }  }}

Use httpUrl (Streamable HTTP). Gemini CLI's url field is SSE.

VS Code (GitHub Copilot)

From VS Code MCP servers and the MCP configuration reference: workspace file .vscode/mcp.json, or MCP: Open User Configuration. VS Code uses a top-level servers object (not mcpServers).

{  "servers": {    "rankreactor": {      "type": "http",      "url": "https://www.rankreactor.ai/mcp"    }  }}

Or run MCP: Add Server from the Command Palette, choose HTTP, and paste the URL. Start the server. When Connect RankReactorAI opens, paste the project key and choose Allow. Then open Copilot Chat in Agent mode and enable the RankReactor tools if asked. If OAuth does not run, the same type: http entry can take headers.Authorization as Bearer <your-project-key> (do not commit a live key).

Perplexity and Le Chat

Perplexity (custom remote connectors): add a custom connector with MCP server URL https://www.rankreactor.ai/mcp, authentication OAuth, transport Streamable HTTP. When Connect RankReactorAI opens, paste the project key and choose Allow.

Mistral Le Chat / Work (MCP connectors): Connectors → Add Connector → Custom MCP Connector, name rankreactor, server URL https://www.rankreactor.ai/mcp. The platform can detect OAuth 2.1 with dynamic client registration. When Connect RankReactorAI opens, paste the project key and choose Allow.

What this repo contains

Client install docs on the product site: https://www.rankreactor.ai/docs/ai-assistants.