Name: Ensures Node.js 22+ and npx for the official DataForSEO MCP server
Brand: Octomind

Ensures Node.js 22+ and npx for the official DataForSEO MCP server

check: npx https://github.com/dataforseo/mcp-server-typescript

Official DataForSEO MCP server for keyword metrics, SERPs, competitor research, and backlink data. The dep ensures Node.js 22+ and npx; the seo-data capability pins the server package.

MCP Server

Package: [email protected]
Transport: stdio (the package's default mode)
Launch: npx -y [email protected]
Runtime: Node.js 22+ with npx, on macOS, Linux, or Windows Git Bash
Source: https://github.com/dataforseo/mcp-server-typescript

The installer delegates platform handling to nodejs/node with a minimum major version of 22 and verifies the result. It does not store credentials or run paid API requests.

Authentication

Get API credentials at https://app.dataforseo.com/api-access. These are an API login and API password, not a single API key or the account password.

Variable	Required	Purpose
`DATAFORSEO_LOGIN`	Yes	DataForSEO API login
`DATAFORSEO_PASSWORD`	Yes	DataForSEO API password

Set both in the environment used to launch Octomind. Keep secret values out of manifests, prompts, command arguments, and git. The capability passes them to the child through env; Octomind withholds it when either is missing. Upstream also accepts DATAFORSEO_USERNAME as a login alias, but this wiring uses DATAFORSEO_LOGIN.

No OAuth is used for stdio. The separately hosted OAuth endpoint is https://mcp.dataforseo.com/v3/mcp; it is not the provider configured here.

Available Tools

Tool	Purpose
`docs_index`	Find API documentation, optionally by section
`docs_list_sections`	List available documentation sections
`docs_search`	Read the documentation for a specific endpoint
`api_request`	Make an authenticated DataForSEO API request

Read endpoint documentation first; use the documented HTTP method, path, and JSON body rather than guessing. api_request uses .ai paths by default; its noAiMode option selects standard paths. Check response and task status codes before treating a result as evidence.

API requests may consume account balance. Confirm the research scope, location, language, sample size, and budget before billable calls; don't run bulk requests or automatic retries without approval. Preserve returned costs, timestamps, and provider/endpoint provenance. A handshake and tools/list do not test credentials, account balance, or data access.

Provider keyword/authority estimates are not Ahrefs DR or observed demand. A provider result is evidence only for its documented surface and capture conditions, not a substitute for a ChatGPT UI run or proof of market-wide AI visibility.

Configuration Example

Agents declare seo-data in their top-level capabilities; this server block belongs only in the capability provider file:

toml

[[mcp.servers]]
name = "dataforseo"
type = "stdio"
command = "npx"
args = ["-y", "[email protected]"]
timeout_seconds = 120
env = { DATAFORSEO_LOGIN = "{{ENV:DATAFORSEO_LOGIN}}", DATAFORSEO_PASSWORD = "{{ENV:DATAFORSEO_PASSWORD}}" }
tools = []

Validate without paid requests:

bash

bash deps/dataforseo/mcp-server.sh
bash scripts/lint-deps.sh deps/dataforseo/mcp-server.sh
bash scripts/lint-capabilities.sh capabilities/seo-data
scripts/mcp-versions.sh test seo-data
bin/load seo:audit
bin/load seo:strategist

View on GitHub