Search Console over MCP
A Google Search Console MCP server you don't have to run.
One line connects your own Search Console to Claude Code, Cursor, Cline or Windsurf. Your agent gets the next move, the page it belongs to and who is ranking above you, instead of rows to sort out.
From $49/mo, everything included · works in Claude Code, Cursor, Cline and Windsurf
The real answer this site got on 2026-09-23. The rivals are the ones Google actually returned.
Two ways to do it
The open-source servers are good. The setup around them is the work.
There are several free Google Search Console MCP servers on GitHub, and they are worth it if you want to run your own. What you take on is a Google Cloud project, an OAuth client whose refresh token expires every 7 days until you publish it (or a service-account key instead), a process to host, and an agent that receives raw rows. QueryScope is the hosted version of the same idea.
Run your own server
Free, and yours to set up and maintain.
Connect QueryScope
One line, hosted, and the call already made.
Four things worth knowing before your agent reads a single row.
If any of the terminology is new, the Search Console glossary defines every metric and indexing state in a sentence, each with the caveat that makes it misleading. If you are weighing writing against the API yourself instead of running a server, the ceilings it imposes are worth reading first.
It is read-only, and it is yours to revoke.
QueryScope reads your Search Console over the webmasters.readonly scope: clicks, impressions, position, the searches each page is shown for, indexing. It can read your data and nothing else, and you can pull the grant in one click from your Google account or from ours. There is no OAuth client of your own to create, publish and keep alive.
scope: webmasters.readonly · you approve · you revoke
How the terminal read compares to the dashboardThe query numbers are already short before your agent sees them.
Google leaves rare searches out of the query report for privacy, so that table always totals less than your real clicks. Ahrefs put the hidden share near half across 22 billion clicks in April 2025. A bare server passes that straight through. QueryScope takes volume from the page level and reads the query list for what a page is shown for, never as a total.
query total ≈ 53% of clicks · page total = 100%
Page-level vs query-level dataGoogle keeps 16 months. Your agent gets longer than that.
Search Console holds a rolling 16-month window and offers no delta export, so the totals you trust today are gone unless something snapshots them forward. On connect QueryScope backfills the full window Google still has, then stores a daily snapshot from that day on. That stored history is also what every measurement is read against later.
backfilled on connect · a daily snapshot kept from then on
The 16-month wall and kept historyYour agent gets the move, and the honest edge of the reading.
You get one target at a time: the search, the page that owns it, which of the seven moves applies, and what Google is returning above you, read from the live results page for the searches your plan declared. Past the click it has nothing to say and says so: a ranking change is never reported as signups or revenue. It never sees your repository, it writes no content, and anything that changes what a visitor reads waits for your commit.
reads: your Search Console, and the results page · stops at: the click
Set up the Google Search Console MCP server in two steps.
Step 1: add the server to your IDE's MCP config with the line below. Step 2: authorize Search Console in the browser that opens, and pick a site.
claude mcp add --transport http queryscope https://mcp.queryscope.dev/mcpWorks the same in Cursor, Cline and Windsurf. Your IDE opens a browser once to authorize.
Then ask: “what should I work on?”
Point your coding agent at your own search data.
Every feature in every plan. Connect over MCP in one line, keep the history Google drops, and get one move at a time.
7-day free trial · nothing due today · cancel in one click
FAQ
Google Search Console over MCP.
Isn't there a free open-source GSC MCP server?
Yes, several, and they are a fine choice if you want to run your own server. QueryScope is the hosted version: no Google Cloud project to set up, a daily snapshot kept past Google's 16-month window, and one next move at a time instead of raw rows. You are paying for the setup you skip and the decision you get.
Do I need my own Google Cloud project or OAuth app?
No. You authorize QueryScope over a read-only Search Console grant you approve and can revoke. There is no OAuth client of your own to create and keep alive, which is the fiddly part of running a server yourself: a client left in testing mode is issued a refresh token that expires every 7 days, so a self-hosted server stops working weekly until you publish it or switch to a service-account key.
Which coding agents and IDEs work?
Claude Code, Cursor, Cline, Windsurf, or any MCP client. It is a standard remote MCP server over Streamable HTTP, so you add one URL to the IDE's MCP config and authorize in the browser once. Your IDE runs the LLM, so there is no separate AI fee. If you are wiring this up in Claude Code, here is how the SEO loop fits together.
Is it the same data as Search Console?
Yes, your own, read over an OAuth grant you approve and can revoke. Plus a daily snapshot kept past Google's 16-month window, so your history stays. Below position 10, where Search Console records nothing, QueryScope reads the live results page instead, for the searches your plan declared. Every position on page 1 comes from Search Console itself.
What can my agent actually do with it?
Four things. Record what your site sells, who buys it and which rivals it competes with, read out of your repo. Size the searches your agent reads out of your repo: how many people search each one, where you stand, and who ranks above you. Record what you decided about each of those searches. Then ask what to work on and get one search, one page, one move, and what to change. Your agent makes the edit; QueryScope never touches your repository.
Is QueryScope connected to Google?
QueryScope is independent and not affiliated with Google. It reads your own Search Console over a read-only grant you approve, and you can revoke it at any time from your Google account or from ours.