Most SEO work inside an AI editor stalls on the same step. You ask the agent what ranks for a query, and it cannot look. So you open a browser, read a results page, copy a few lines back into the chat, and the agent reasons over your transcription instead of the page.

The fix is not a better prompt. It is giving the agent a tool that fetches the results page itself. Model Context Protocol is the wire format for that, and an SEO MCP server is simply one that exposes search tasks as callable tools.

What an SEO MCP server actually is

It is a server that turns SEO tasks into tools an agent can call. Instead of you reading a results page and pasting the ranks into a prompt, the agent asks for the ranks. The difference matters most in a loop: an agent that can capture a SERP, read it, rewrite a title and capture again is doing the work, not narrating it.

Nearly every SEO MCP server on the market is a protocol wrapper over a metered data API. That is a real architecture with real advantages, and it is why those servers can answer questions about search volume that this one cannot.

Start with what this does not do. There is no search volume here, no keyword difficulty, no cost per click, no backlink index, no domain overview, no stored rank history and no site crawler. If you need any of those, a vendor backed suite such as OpenSEO has them and ToolPiper does not. OpenSEO is open source under MIT with nine agent skills, and its data is metered through DataForSEO on every keyword, SERP and backlink call. Verified 2026-08-28.

What ToolPiper does instead

ToolPiper captures the live results page in the Chrome already installed on your Mac, over the DevTools Protocol. Because it reads the rendered page rather than a vendor feed, the capture includes the People Also Ask block, the related searches, and on Google the AI Overview together with the domain it cites first. OpenSEO's SERP tool flattens every result to six fields, so an AI Overview arrives as a typed row with the contents dropped.

The trade is legible. You give up volume and history. You get the answer box, its sources, and a flat price with no metering: no API key, no data vendor, no project, no credits. OpenSEO's MCP asks for a project on every call. This one asks for a query.

What you need

A Mac on Apple Silicon running macOS 26 or later, Google Chrome, ToolPiper, and any MCP client. Claude Code is the one used below; Cursor, Devin and Zed take the same server.

Connect the server, then ask

Once ToolPiper is registered as an MCP server, the SEO tools appear alongside every other tool the agent has. You do not call them by name. You ask for the outcome and the agent picks the tool.

Limits worth knowing before you start

ToolPiper needs macOS 26 or later on Apple Silicon, and the SERP capture drives Google Chrome over the DevTools Protocol, so Chrome has to be installed. One capture is a sample, not a measurement: AI Overview presence rotates by session, location and personalisation, so a single run tells you what happened once, in your browser, from where you are sitting. Search Console tools sit behind Google OAuth on your own account.

Try it

Download ToolPiper, add it to Claude Code with one command, and ask your agent who owns the AI Overview on a query you care about. The rung badge at the top of this page names the plan the SEO tools sit on, resolved from the pricing catalog rather than written here.

Part of the SEO tools guide for Mac. The roundup covers SERP capture, keyword discovery, rank checks, Search Console and listening in one place.