DataForSEO Brain

DataForSEO MCP Server (official)

4 min read

DataForSEO MCP Server (official)

The official TypeScript/Node Model Context Protocol server that lets AI agents call DataForSEO APIs through one standardized interface. Sits under DataForSEO Brain → Entities.

Overview

The DataForSEO MCP server is a TypeScript / Node.js implementation of Anthropic’s Model Context Protocol that lets AI assistants call selected DataForSEO APIs and obtain SEO data through a single standardized interface. It is published by DataForSEO under the Apache-2.0 license (~221 stars / ~120 forks on GitHub as of mid-2026). The framing is that MCP removes per-source custom integration work: it uses a JSON-RPC client-server architecture so non-technical users can request premium SEO data via natural-language prompts instead of hand-coded API calls.

Modules / tools exposed

The server is organized into modules, each wrapping an API family and exposing multiple tools (from the README, the authoritative superset):

  • SERP - real-time Google, Bing, Yahoo results
  • KEYWORDS_DATA - search volume, CPC, clickstream keyword metrics
  • ONPAGE - website crawling and on-page performance
  • DATAFORSEO_LABS - proprietary keyword/SERP/domain databases and algorithms
  • BACKLINKS - inbound links and referring-domain analysis
  • BUSINESS_DATA - reviews/listings (Google, Trustpilot, Tripadvisor)
  • DOMAIN_ANALYTICS - technology-stack detection and Whois
  • CONTENT_ANALYSIS - citation tracking and sentiment
  • AI_OPTIMIZATION - keyword discovery, conversational optimization, LLM benchmarking
  • MERCHANT - product/pricing/seller data (Google Shopping, Amazon)

By default the server registers hundreds of tools, which consumes significant LLM context; modules are built on a BaseModule class (getTools()), registered in modules.config.ts, so operators are expected to narrow the surface.

Configuration (env vars)

  • DATAFORSEO_USERNAME - required (API login)
  • DATAFORSEO_PASSWORD - required (API password, not the dashboard password - see cap-authentication-security)
  • ENABLED_MODULES - comma-separated allow-list (e.g. "SERP,KEYWORDS_DATA,ONPAGE"); the primary permission gate. If unset, all modules load.
  • ENABLED_PROMPTS - allow-list of prompt templates (e.g. top_3_google_result_domains)
  • DATAFORSEO_FULL_RESPONSE - true returns unfiltered raw responses (default false)
  • DATAFORSEO_SIMPLE_FILTER - true flattens the filter schema for LLMs that struggle with nesting
  • A JSON field-config.json (supported_fields map keyed by tool) further trims returned fields, passed with --configuration field-config.json.

Transports & distribution

  • stdio (default) - npx dataforseo-mcp-server (the mode Claude Desktop uses)
  • Streamable HTTP - npx dataforseo-mcp-server http (port 3000; not Claude Desktop)
  • SSE - legacy/deprecated
  • Hosted remote - DataForSEO runs https://mcp.dataforseo.com/mcp (streamable HTTP) and https://mcp.dataforseo.com/http (HTTP/SSE), so no self-hosting is required
  • Cloudflare Worker - index-worker.js build with POST /mcp, GET /sse, GET /health, deployable via wrangler
  • Published on npm as dataforseo-mcp-server (latest 2.9.9 at retrieval, Jun 2026; requires Node.js >=20.0.0 per package.json engines - the README still says v14+ and the MCP page 18+, both stale). The v2.9.9 line added Historical SERPs (date params), an exclude_targets field on the backlinks domain-intersection tool, tool annotations/titles, and OAuth/MCP auth improvements. Documented clients: Claude Desktop, Claude Code, Cursor, ChatGPT, Gemini CLI, Docker.

When to use / how it fits

Gotchas / limits

  • Loading all modules drowns the agent in tool definitions and tokens; always set ENABLED_MODULES to shrink both attack surface and token cost.
  • Marketing/help pages sometimes list only the “seven major” modules, omitting AI_OPTIMIZATION and MERCHANT; the README is the superset (documentation lag, not a contradiction).
  • Many MCP paths assume a per-user interactive session and lack a clean static-credential, server-to-server auth option - a reason to prefer REST for unattended jobs.

Sources

Graph View

Backlinks