SEO

DataForSEO MCP: setup guide and a hosted alternative

The DataForSEO MCP server connects AI agents to SEO data: keywords, SERP results, backlinks, and technical crawl. Here is how to set it up in Claude, ChatGPT, and Cursor, fix common problems, and when to use a hosted alternative instead.

Jesse Sibley
By , FounderPublished · Last updated · 11 min read

DataForSEO MCP is an MCP server that connects AI agents to the DataForSEO API for SEO data: keywords, SERP results, backlinks, domain metrics, technical crawl, and AI optimization data. The official server runs hosted at https://mcp.dataforseo.com/v3/mcp or locally via npx dataforseo-mcp-server. You pay DataForSEO per API call with a $50 minimum payment, and the balance never expires. The MCP server itself is free.

This guide walks through setting up the official DataForSEO MCP server in Claude, ChatGPT, and Cursor, covers common authentication and schema problems, shows real example prompts, and explains when a hosted alternative like Manifold fits better.

What is the DataForSEO MCP server?

The DataForSEO MCP server is a bridge between your AI agent and DataForSEO's API. It exposes DataForSEO's SEO data endpoints as MCP tools your agent can call. The server handles authentication, request formatting, and response parsing, so your agent can ask for SEO data in plain language and get structured results back.

The server includes tools for:

  • Keywords: Search keywords, get metrics (volume, KD, CPC), find related terms
  • SERP: Live Google results for any keyword, location, and device
  • Backlinks: Domain backlink profiles, referring domains, anchor text
  • Domain metrics: Domain rank (authority score), organic traffic estimates, ranked keywords
  • Technical crawl: Site audits, on-page factors, technical issues
  • AI optimization: ChatGPT, Claude, Gemini, and Perplexity prompt tracking
  • Business data: Reviews, local listings, and business information

DataForSEO sources its data from its own crawlers and indexes, not from third-party providers. The MCP server is open source on GitHub at dataforseo/mcp-server-typescript.

Setup: hosted remote server

The fastest way to connect is via the hosted remote server at https://mcp.dataforseo.com/v3/mcp. It works in Claude, ChatGPT, and Cursor. You need a DataForSEO account with API credentials.

Step 1: Get your API credentials

Sign up at dataforseo.com if you do not have an account. DataForSEO offers a $1 trial credit, then requires a $50 minimum payment. Your balance never expires.

Go to the API Access dashboard and copy your API login and API password. These are different from your dashboard login credentials. The API credentials look like your_api_login and your_api_password, not your email address.

For remote HTTP connections that use Basic auth, you need a Base64-encoded token. The API Access page has a "Send by Email" button that delivers a pre-encoded token. Alternatively, encode your_api_login:your_api_password in Base64 yourself.

Step 2: Connect in Claude

Open Claude Desktop, go to Customize → Connectors, click "Add custom connector", and paste https://mcp.dataforseo.com/v3/mcp. Approve the OAuth request when prompted.

Alternatively, edit your config file (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json, Windows: %APPDATA%\Claude\claude_desktop_config.json):

JSON
{
  "mcpServers": {
    "dataforseo": {
      "command": "npx",
      "args": ["-y", "dataforseo-mcp-server"],
      "env": {
        "DATAFORSEO_USERNAME": "your_api_login",
        "DATAFORSEO_PASSWORD": "your_api_password"
      }
    }
  }
}

Step 3: Connect in ChatGPT

  1. Log into your OpenAI API account and go to the Dashboard
  2. Create a new chat with your preferred ChatGPT model
  3. Click Tools → MCP Server → Add new
  4. Paste the URL: https://mcp.dataforseo.com/v3/mcp
  5. Add a custom label (e.g., "DataForSEO")
  6. For authentication, choose "Custom headers"
  7. In the Header field, write Authorization
  8. In the Value field, write Basic followed by your Base64-encoded API credentials (get pre-encoded credentials via the "Send by Email" button in your DataForSEO API Access dashboard)
  9. Click Connect and configure approval settings as needed

Step 4: Connect in Cursor

Open Settings → Tools & Integrations → MCP Tools, click "Add Custom MCP", and paste this (replace your Base64 token):

JSON
{
  "mcpServers": {
    "dataforseo": {
      "url": "https://mcp.dataforseo.com/v3/mcp",
      "headers": {
        "Authorization": "Basic <your_base64_token>"
      }
    }
  }
}

Note: Cursor limits you to 40 active MCP tools. Disable unused modules if you hit the limit.

Setup: local server

Run the server locally via npx -y dataforseo-mcp-server (requires Node.js 18+). Then configure your client with environment variables DATAFORSEO_USERNAME and DATAFORSEO_PASSWORD in the MCP config. See the DataForSEO setup guide for client-specific config examples.

Common problems and fixes

Authentication errors

Error: "You are not authorized to access this resource"

Cause: Wrong credentials or wrong credential format.

Fix:

  1. Use your API credentials (from the API Access dashboard), not your dashboard login
  2. Check the format your client expects:
    • Remote HTTP with Basic auth: Base64-encoded login:password
    • Local stdio: Plain DATAFORSEO_USERNAME and DATAFORSEO_PASSWORD environment variables
    • OAuth: Approve the authorization request when prompted

Schema validation errors

Error: "Invalid schema for function" or "array schema missing items"

Cause: Some AI clients do not fully support DataForSEO's OpenAPI schema format, especially for complex Labs endpoints that accept filters or order_by parameters.

Fix:

Self-host the MCP server and enable simplified schema mode:

Terminal
export DATAFORSEO_SIMPLE_FILTER=true
npx dataforseo-mcp-server

Point your client to the local server (usually http://localhost:3000 for HTTP mode, or use stdio config).

Tools commonly affected:

  • dataforseo_labs_google_ranked_keywords
  • AI optimization endpoints (with filters)
  • Trends endpoints (with complex nested parameters)

Token limit errors

Error: Response is too large or truncated

Cause: DataForSEO returns comprehensive data. Some endpoints return hundreds of results per call, and the agent's context window fills up.

Fix:

  1. Limit results with pagination parameters (limit and offset)
  2. Ask your agent to fetch only specific fields you need
  3. Use narrower queries (target specific domains, keywords, or date ranges)

Too many tools

Error: Cursor shows "40 active tools limit reached"

Cause: Cursor limits you to 40 MCP tools. DataForSEO exposes over 40 endpoints.

Fix:

Disable other MCP servers or modules you do not need. Each DataForSEO endpoint becomes a separate tool in the client.

"Fetch not defined" error

Error: "Fetch is not defined"

Cause: Old Node.js version.

Fix: Upgrade to Node.js 18 or higher.

Example prompts

Here are real prompts you can use with the DataForSEO MCP server once connected.

Keyword research

"Find 20 keywords related to 'crm for startups' with at least 200 monthly searches in the US. Show volume, KD, and CPC for each."

Rank check

"Check where example.com ranks on Google US desktop for 'project management software' and 'best project management tools'."

"Show the top 20 referring domains linking to competitor.com, sorted by domain rank. Include anchor text distribution."

Domain overview

"Analyze example.com: domain rank, estimated organic traffic, number of ranked keywords, and top 10 keywords by traffic."

Technical audit

"Crawl example.com and report broken links, missing meta descriptions, slow pages, and duplicate content issues."

AI visibility

"Check whether ChatGPT, Claude, and Gemini recommend my brand for 'best CRM for agencies'. List competitors they mention and which sources they cite."

Run this in your agent

Connect Manifold to Claude, ChatGPT or Cursor, then paste this prompt.

Find keywords where we rank 5 to 15 that could reach page one. For each, show who holds the top three spots and what their pages cover.

When to use Manifold instead

DataForSEO MCP is a strong choice when you already have a DataForSEO account, understand API pricing, and want raw access to their full endpoint catalog. But it has trade-offs:

  • You manage your own DataForSEO account and API credentials
  • You configure the MCP server yourself (hosted or local)
  • You pay DataForSEO per API call (minimum $50, no monthly plan)
  • Schema validation issues require self-hosting with simplified mode
  • No Search Console or Bing data (DataForSEO only)

Manifold is a hosted alternative that runs DataForSEO data plus other marketing data sources behind one server. It fits better when:

  • You want one hosted server with OAuth 2.1, no API keys of your own to manage
  • You want per-call credits instead of a DataForSEO minimum payment upfront
  • You need Search Console and Bing data alongside SEO (0 credits per call)
  • You want social, Reddit, ad library, and lead data in the same server
  • You prefer not to debug authentication or schema validation yourself

Here is how they compare:

FeatureDataForSEO MCP (official)Manifold
Data sourceDataForSEO APIDataForSEO plus other marketing data sources
Hosted or localBoth (remote at mcp.dataforseo.com or local npx)Hosted only (mcp.manifoldmcp.com)
AuthenticationOAuth (Claude, ChatGPT) or Basic auth with API credentialsOAuth 2.1 or API key
Entry cost$50 minimum (DataForSEO), $1 trial500 free credits per workspace, top-ups from $10
Pricing modelPer API call, balance never expiresPer call with credits, or Pro from $20/mo
SetupConfigure MCP client with credentials or OAuthConnect via OAuth, no API keys of your own
Tools exposed40+ DataForSEO API endpoints88 tools: SEO, AI visibility, Search Console (0 credits), Bing (0 credits), social, Reddit, ads, leads
Schema issuesSelf-host with DATAFORSEO_SIMPLE_FILTER=trueHandled by Manifold
Search ConsoleNoYes (0 credits per call)
Bing WebmasterNoYes (0 credits per call)
AI answersYes (ChatGPT, Claude, Gemini, Perplexity)Yes (same engines)
Social and adsNoYes (TikTok, Instagram, Facebook, YouTube, LinkedIn, X, Reddit, Meta Ads, Google Ads)
LeadsBusiness data onlyCompany search, person search, email finder, email verification
Best forTeams who want raw DataForSEO API access and can manage credentialsTeams who want one hosted server across SEO, AI visibility, social, ads, and leads with per-call pricing

Both servers run on DataForSEO data for SEO, so keyword metrics, SERP results, backlinks, and domain rankings are the same. The difference is in setup complexity, pricing structure, and what else you get in the same server.

Questions people ask

Is the DataForSEO MCP server free?

The MCP server itself is free and open source. You pay DataForSEO for API calls. The minimum payment is $50, and your balance never expires. There is a $1 trial credit to test the API.

Do I need a DataForSEO account to use the official server?

Yes. The DataForSEO MCP server connects to DataForSEO's API, so you need API credentials from a DataForSEO account. Sign up at dataforseo.com.

What is the difference between API credentials and dashboard login?

Your API credentials (API login and API password) are for making API calls. They are different from the email and password you use to log into the DataForSEO dashboard. Find your API credentials in the API Access dashboard.

Can I use the remote server without creating a DataForSEO account?

No. The remote server at https://mcp.dataforseo.com/v3/mcp authenticates with your DataForSEO API credentials via OAuth or Basic auth. If you do not want to create a DataForSEO account, use Manifold instead, which handles DataForSEO data behind one hosted server with OAuth 2.1.

Which MCP clients work with DataForSEO?

The official DataForSEO MCP server works in any MCP-compatible client: Claude Desktop, Claude Code, ChatGPT (via plugins), Cursor, Gemini CLI, Zed, Cline, and others. OAuth is supported in Claude and ChatGPT. Other clients use Basic auth (remote) or environment variables (local).

How do I fix "you are not authorized" errors?

Use your API credentials (from the API Access dashboard), not your dashboard login. For remote HTTP, encode your credentials as Base64 login:password. For local stdio, set DATAFORSEO_USERNAME and DATAFORSEO_PASSWORD environment variables.

Why do some tools fail with schema validation errors?

Some AI clients do not fully support DataForSEO's OpenAPI schema, especially for Labs endpoints with filters or order_by. Self-host the server and set DATAFORSEO_SIMPLE_FILTER=true to enable simplified schema mode, then point your client to the local server.

How much does DataForSEO cost?

You pay per API call. Prices vary by endpoint. The minimum payment is $50, and your balance never expires. Check dataforseo.com/pricing for detailed pricing by endpoint.

Does it include Search Console or Bing data?

No. DataForSEO MCP only covers DataForSEO's own API, which does not include Search Console or Bing Webmaster Tools. If you need Search Console and Bing data, use Manifold, which includes both at 0 credits per call.

Can I run it in Docker?

Yes. DataForSEO publishes an official Docker image at dataforseo/mcp:latest. Run it with:

Terminal
docker run -p 3000:3000 \
  -e DATAFORSEO_USERNAME=your_api_login \
  -e DATAFORSEO_PASSWORD=your_api_password \
  dataforseo/mcp:latest

The MCP server will be available at http://localhost:3000.

Questions people ask

Is the DataForSEO MCP server free?

The MCP server itself is free and open source. You pay DataForSEO for API calls. The minimum payment is $50, and your balance never expires. There is a $1 trial credit.

Do I need a DataForSEO account?

Yes. The DataForSEO MCP server connects to DataForSEO's API, so you need API credentials from a DataForSEO account. Sign up at dataforseo.com.

What is the difference between API credentials and dashboard login?

Your API credentials (API login and API password) are for making API calls. They are different from the email and password you use to log into the DataForSEO dashboard. Find your API credentials in the API Access dashboard.

How do I fix authentication errors?

Use your API credentials from the API Access dashboard, not your dashboard login. For remote HTTP, encode credentials as Base64 login:password. For local stdio, set DATAFORSEO_USERNAME and DATAFORSEO_PASSWORD environment variables.

Why do some tools fail with schema validation errors?

Some AI clients do not fully support DataForSEO's OpenAPI schema, especially for Labs endpoints with filters or order_by. Self-host the server and set DATAFORSEO_SIMPLE_FILTER=true to enable simplified schema mode.

Does it include Search Console or Bing?

No. DataForSEO MCP only covers DataForSEO's own API. If you need Search Console and Bing, use Manifold, which includes both at 0 credits per call.

Sources

Jesse Sibley

Jesse Sibley

Founder of Manifold. A growth engineer who has worked with London startups for eight years.

All SEO guides
  • 9 min read

    The best SEO MCP servers, compared

    When your agent needs SEO data, an MCP server connects it to the tools you already use. Here are the best SEO MCP servers, what they cost, how they work, and which one fits your workflow.

  • 5 min read

    The best Google Search Console MCP servers

    Google has no official Search Console MCP server. Community-built servers are free but need a Google Cloud project. Hosted options skip that setup. Here are the best Google Search Console MCP servers and which one fits your workflow.

  • 6 min read

    The best MCP servers for marketing teams

    Marketing teams juggle SEO, paid ads, CRM, analytics, social listening, and outbound campaigns. Here are the best MCP servers to connect your AI agent to marketing data, organized by what they do and how they charge.