Is the Mangools MCP Server stable?

It's currently in beta. The set of available tools and their parameters can change as we iterate based on feedback. We'll communicate breaking changes ahead of time on this page.

Which AI clients does it work with?

Any MCP-compatible client. We've verified it with Claude (web, desktop and mobile), ChatGPT, Cursor and Claude Code. Other clients that support remote MCP servers over HTTP should work as well.

Does it consume my Mangools quota?

Yes. Every tool call goes through the Mangools API and counts against your plan's API quota, just like a normal API call. It doesn't matter whether you connect by signing in with your Mangools account or with your API token — both draw from the same daily limits. Keyword API plans have their own dedicated keyword lookup quota; regular plans use their standard API request limits.

Can my AI accidentally spend paid resources?

A few tools (such as creating SERPWatcher trackings or AI Search Watcher monitors) consume paid quota. They're clearly marked as paid in the tool description, and your AI client should ask you for confirmation before calling them. For maximum control, use a client that requires explicit approval for tool calls.

Is my data sent to OpenAI or Anthropic?

The MCP server only talks to your AI client. Whatever your client passes to the underlying model provider (Anthropic, OpenAI, etc.) is controlled by that client, not by Mangools. Refer to your client's privacy policy for details.

Can I use the MCP server with my own integrations?

The MCP server is designed for MCP-compatible AI clients. For custom integrations and scripts, use the regular Mangools REST API with your API token from the API token page.

Do I still need an API token?

Only if your client can't do the OAuth sign-in, or if you use the Mangools REST API directly. For Claude, ChatGPT, Cursor and Claude Code you just sign in.

I see an error in my AI client — what now?

If you connected via sign-in and calls start failing with authorization errors, disconnect and authorize again in your client. If the sign-in window never opens, update your client — several added or fixed MCP OAuth support in recent releases. If you authenticate with a token, double-check that the server URL is exact, that the x-access-token header contains a valid token (no quotes, no extra whitespace), and that your client is restarted after the configuration change. If the error persists, contact our support with the client name and the exact error message.

BETA

Mangools MCP Server

Connect Claude, Cursor and other AI assistants directly to Mangools SEO data — keyword research, SERP analysis, rank tracking and backlinks, right from your AI chat.

Connect your AI

What is MCP?

MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external data sources.
With the Mangools MCP Server, Claude or Cursor can call Mangools tools as if they were power users.
Ask your AI to research keywords, monitor SERPs or audit backlinks — and let it run the queries through Mangools automatically.

What you can do with it

Keyword research from chat

Ask Claude to find keyword ideas, search volumes and SEO difficulty for your topic — without ever leaving the conversation.

Automated SERP analysis

Let an AI agent monitor SERP results, compare competitors and surface ranking opportunities for the queries you care about.

Rank & backlink monitoring

Track positions in SERPWatcher and audit backlinks in LinkMiner through natural-language prompts and AI-generated reports.

Quick start in 3 steps

Choose your AI client

Claude (web, desktop, mobile), ChatGPT, Claude Code, Cursor and other MCP-compatible clients all work.

Add the Mangools server

Paste the server URL into your client — as a custom connector in Claude or ChatGPT, or in the MCP config for other clients.

See snippets

Sign in & authorize

Your client opens the Mangools sign-in in the browser. Log in, approve access, and the tools are ready. No token copying.

Your client doesn’t support OAuth sign-in? Use your API token instead — see Authentication below.

Configuration

Pick your AI client below, copy the snippet, paste it in.

Server URL https://mcp.mangools.com/mcp

In Claude, the Mangools server is added as a custom connector — no config file, no JSON:

Open Settings → Connectors (claude.ai/settings/connectors) and click Add custom connector.
Name it Mangools and paste the server URL: https://mcp.mangools.com/mcp
Click Add, then Connect — and sign in with your Mangools account.

Set it up once on claude.ai or in Claude Desktop — connectors are tied to your Claude account, so it also works in the Claude mobile apps.

Custom connectors are available on every Claude plan — Free accounts can add one custom connector.

On Team or Enterprise plans a workspace Owner adds the connector under Organization settings → Connectors, then each member clicks Connect and signs in with their own Mangools account.

Advanced: config-file setup via mcp-remote

Claude Desktop’s config file only speaks stdio, so a remote HTTP server is reached through the mcp-remote bridge (requires Node.js). Edit ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or %APPDATA%\Claude\claude_desktop_config.json on Windows, and add this to the mcpServers object:

claude_desktop_config.json

{
  "mcpServers": {
    "mangools": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://mcp.mangools.com/mcp"
      ]
    }
  }
}

mcp-remote runs the OAuth sign-in itself and opens your browser — if it can’t open one, it prints the sign-in URL to copy.

Restart Claude Desktop after saving the file — the first launch may take a few extra seconds while mcp-remote downloads.

If the server is skipped on launch, Claude Desktop probably can’t find npx on its minimal PATH (common with nvm). Replace "npx" with the absolute path — run which npx in your terminal and paste the result.

In ChatGPT, the Mangools server is added as a connector — no config file, no JSON:

Open Settings → Connectors → Add MCP server (Developer mode required).
Name it Mangools and paste the server URL: https://mcp.mangools.com/mcp
Sign in with your Mangools account when prompted.

Add this to .cursor/mcp.json in your project, or to ~/.cursor/mcp.json to use it in every project:

mcp.json

{
  "mcpServers": {
    "mangools": {
      "url": "https://mcp.mangools.com/mcp"
    }
  }
}

In Cursor Settings → Tools & MCP the server shows as needing authentication — click Connect and finish the sign-in in the browser.

Run this one-liner from your terminal:

terminal

claude mcp add --transport http mangools https://mcp.mangools.com/mcp

Then run /mcp inside Claude Code and pick Authenticate — your browser opens the Mangools sign-in. Or run claude mcp login mangools from the shell.

In Microsoft Copilot Studio, open your agent and go to Tools → Add a tool → New tool → Model Context Protocol. In the wizard, fill in these fields:

MCP onboarding wizard

Server name:     mangools
Server URL:      https://mcp.mangools.com/mcp
Authentication:  OAuth 2.0 (Dynamic discovery)

With Dynamic discovery, Copilot Studio registers itself with the server automatically. Sign in with your Mangools account when creating the connection, then publish the agent so the tools become available. Requires the Streamable HTTP transport — which is what we serve.

Fallback: API key header (older versions)

MCP onboarding wizard

Server name:          mangools
Server URL:           https://mcp.mangools.com/mcp
Authentication type:  API key
  Type:               Header
  Header name:        x-access-token
API key value:        YOUR_API_TOKEN  (entered when creating the connection)

The header name is a free-text field — type x-access-token. Your token is supplied when you create the connection, then publish the agent so the tools become available.

Replace YOUR_API_TOKEN with your token from the API token page.

In n8n, add the MCP Client Tool node to your AI agent workflow and configure it:

MCP Client Tool node

Endpoint URL:       https://mcp.mangools.com/mcp
Server Transport:   HTTP Streamable
Authentication:     OAuth2
  Use Dynamic Client Registration:  On  (default)
Tools to include:   All

OAuth2 with Dynamic Client Registration needs n8n 1.119.0 or newer. Connect the MCP Client Tool node to your AI Agent node’s tool input. Works with any chat model — OpenAI, Anthropic, local Ollama, etc.

Fallback: Header Auth (older self-hosted instances)

MCP Client Tool node

Endpoint URL:       https://mcp.mangools.com/mcp
Server Transport:   HTTP Streamable
Authentication:     Header Auth
  Header name:      x-access-token
  Header value:     YOUR_API_TOKEN
Tools to include:   All

Replace YOUR_API_TOKEN with your token from the API token page.

Native MCP support requires Open WebUI v0.6.31 or newer. Go to Admin Panel → Settings → External Tools → + and fill in the form:

Add Connection form

Type:  MCP (Streamable HTTP)
Name:  mangools
URL:   https://mcp.mangools.com/mcp
Auth:  OAuth 2.1 (Dynamic Client Registration)

Make sure WEBUI_SECRET_KEY is set on your instance — without it, OAuth tokens are lost on container restart. Only the Streamable HTTP transport is supported natively.

Fallback: token in a header (older installs)

Add Connection form

Type:     MCP (Streamable HTTP)
Name:     mangools
URL:      https://mcp.mangools.com/mcp
Auth:     None
Headers:  { "x-access-token": "YOUR_API_TOKEN" }

Leave Auth on None — the token goes in the free-form Headers field as a JSON object.

Replace YOUR_API_TOKEN with your token from the API token page.

Available tools

Around 35 tools across all Mangools products.
Click each product to see what your AI can do.

KWFinder — keyword research

Search related keywords with volume, CPC and SEO difficulty
Get SERP results for any keyword
Find competitor domains for a keyword
Get competitor keywords for a domain
Get keyword suggestions for a URL
Compute SEO difficulty for any URL
Manage keyword lists (create, rename, add keywords)
Batch keyword check (up to 700 keywords per request)
Content gap analysis between domains
View quota usage and history

SERPChecker — SERP analysis

Fetch SERP results for any keyword and location
Get Moz and Majestic URL metrics
List historical SERPChecker requests
Retrieve SERP snapshots over time

SERPWatcher — rank tracking

List all of your trackings
Get tracking detail and stats over time
List tracked keywords and their positions
Create new trackings (single or batch)
Update tracking settings and tracked keywords
Manage reports, annotations and tags

Creating trackings and adding tracked keywords may consume paid quota depending on your plan.

LinkMiner — backlink analysis

Get URL metrics (DA, PA, citation flow, trust flow)
Fetch backlinks for any domain
Manage favorite links
List historical LinkMiner requests and exports

SiteProfiler — website analysis

Get a domain’s overview metrics
Inspect audience and top content
Pull the backlink profile
List a domain’s top competitors
List historical SiteProfiler requests

AI Search Watcher — LLM visibility

List supported AI models and monitors
Get a monitor’s detail, prompts and settings
Generate prompts for a brand or topic
Create new monitors and add prompts
Update monitor settings

Creating monitors consumes paid quota.

Utilities

Search Mangools-supported locations (countries, cities)
Get details for a specific location

Authentication

Sign in with your Mangools account (recommended). The MCP server supports the standard MCP OAuth flow. Your AI client redirects you to the Mangools sign-in, you approve access, and you’re connected — no tokens to copy, nothing to paste into config files.

Losing access? If tool calls start failing with authorization errors, disconnect the server in your client and authorize again.

API token still works. Every client that can send headers can keep authenticating with your API token as an x-access-token header — and it’s what the Mangools REST API uses. Find it on the API token page — every user has one enabled by default.

How do I rotate it? Reset your API key on the same page. The old token stops working immediately, so update your MCP client configuration right after.

Treat the token like a password — anyone with access can run Mangools queries on your behalf and consume your quota.

Frequently Asked Questions

Is the Mangools MCP Server stable?: It’s currently in beta. The set of available tools and their parameters can change as we iterate based on feedback. We’ll communicate breaking changes ahead of time on this page.
Which AI clients does it work with?: Any MCP-compatible client. We’ve verified it with Claude (web, desktop and mobile), ChatGPT, Cursor and Claude Code. Other clients that support remote MCP servers over HTTP should work as well.
Does it consume my Mangools quota?: Yes. Every tool call goes through the Mangools API and counts against your plan’s API quota, just like a normal API call. It doesn’t matter whether you connect by signing in with your Mangools account or with your API token — both draw from the same daily limits. Keyword API plans have their own dedicated keyword lookup quota; regular plans use their standard API request limits.
Can my AI accidentally spend paid resources?: A few tools (such as creating SERPWatcher trackings or AI Search Watcher monitors) consume paid quota. They’re clearly marked as paid in the tool description, and your AI client should ask you for confirmation before calling them. For maximum control, use a client that requires explicit approval for tool calls.
Is my data sent to OpenAI or Anthropic?: The MCP server only talks to your AI client. Whatever your client passes to the underlying model provider (Anthropic, OpenAI, etc.) is controlled by that client, not by Mangools. Refer to your client’s privacy policy for details.
Can I use the MCP server with my own integrations?: The MCP server is designed for MCP-compatible AI clients. For custom integrations and scripts, use the regular Mangools REST API with your API token from the API token page.
Do I still need an API token?: Only if your client can’t do the OAuth sign-in, or if you use the Mangools REST API directly. For Claude, ChatGPT, Cursor and Claude Code you just sign in.
I see an error in my AI client — what now?: If you connected via sign-in and calls start failing with authorization errors, disconnect and authorize again in your client. If the sign-in window never opens, update your client — several added or fixed MCP OAuth support in recent releases. If you authenticate with a token, double-check that the server URL is exact, that the x-access-token header contains a valid token (no quotes, no extra whitespace), and that your client is restarted after the configuration change. If the error persists, contact our support with the client name and the exact error message.

Connect your AI