Google Search Console MCP Server
Connect Claude, ChatGPT or any AI tool to your Search Console properties. Ask which queries are losing clicks, why a page dropped out of the index, or whether your sitemap actually got read — in plain English, against live data.
What is the Google Search Console MCP?
The Pipeboard Google Search Console MCP (Model Context Protocol) Server connects AI assistants like Claude and ChatGPT to the Search Console API. Instead of filtering the Performance report by hand or exporting rows into a spreadsheet, you ask a question and the assistant queries your properties directly.
It is built for SEO practitioners, marketers and agencies who want answers about organic search — queries, pages, countries, devices, index coverage and sitemap health — without clicking through the interface for each one.
What you can do with it:
Paid and organic in the same conversation
Search Console on its own answers organic questions. What it cannot tell you is whether you are paying for clicks you were already winning for free, or whether a term you rank nowhere for is worth bidding on. Those questions need both sides of the search page at once.
If you also connect Google Ads, the same assistant can read your search terms report and your Search Console queries in one session — so it can answer them without you exporting anything. The unified Ads MCP endpoint carries both through a single connection.
Brand overlap
"Which of my Google Ads search terms are queries I already rank in the top 3 for organically?"
Coverage gaps
"Find queries with high impressions but position worse than 10 in Search Console, and tell me which ones I am not bidding on."
Cannibalisation
"Compare organic clicks and paid clicks for my top 20 non-brand queries over the last 90 days."
Hosted or self-hosted: which one you want
There are good open-source Search Console MCP servers on GitHub, and the Search Console API itself is free. If you are comfortable running a local server and managing your own OAuth client, a self-hosted one will do the reading job well. Here is the honest difference.
Self-hosted, open source
- Free, and the code is yours to read and change
- You create a Google Cloud project, enable the API and manage OAuth credentials
- Runs on one machine — reconfigure it on each device you use
- Usually one Google account at a time
- Access control is whatever the Google account can reach
Pipeboard, hosted
- No Google Cloud project — click Connect and approve Google’s consent screen
- One URL that works from Claude, ChatGPT and any MCP client, on any device
- Several Google accounts at once, with discovery aggregated across them
- Per-property access control, so a connection reaches only the properties you pick
- Read-only by default; write access is a separate, explicit grant
- Token refresh handled for you, and Search Console alongside your ad platforms
How Pipeboard uses your Search Console data
What we request, why, and what we never do with it
When you connect Google Search Console, Pipeboard requests access through Google’s official OAuth consent screen. We request only the scopes needed for the tools described on this page, plus your basic account identity so we can label which Google account a property belongs to. We use this access solely to fulfil the requests you (or your AI assistant, on your behalf) make.
The access we request, and why:
webmasters.readonlyThe default for every new connection. Runs performance queries against your properties, inspects URL index status, and lists the properties and sitemaps you can reach. This read scope powers every reporting tool.
webmastersNot requested by default. Only asked for when you explicitly turn on write access from your connections page — Google has no narrower scope for it. It is what lets the assistant submit or delete a sitemap and add or remove a property. It never alters your historical performance data, which Search Console does not allow in any case.
userinfo.emailIdentifies the connected account so we can label which Google login a property belongs to. Important if you connect more than one Google account.
userinfo.profileDisplays the connected account name in your Pipeboard connections list so you can tell your connections apart.
Our commitments
- We never sell your Search Console data, and we never use it for advertising.
- Your data is used only to respond to the requests you or your AI assistant make — it is not used to train AI models.
- You can disconnect Google Search Console at any time from your connections page, which revokes Pipeboard’s access.
- Read-only by default: the write scope is requested only if you turn write access on yourself.
- Pipeboard’s use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.
Full details are in our Privacy Policy and Security pages.
Getting Started
Set up in under 2 minutes
1. Sign up for a Pipeboard account:
pipeboard.co/auth/signup →2. Connect Google Search Console (you’ll review and approve the read-only access above on Google’s consent screen):
pipeboard.co/connections →3. Add the MCP server to Claude (or another MCP client):
https://google-search-console.mcp.pipeboard.co/Use this URL as a Streamable HTTP MCP server in your client. Authentication is handled automatically through your Pipeboard session.
Connect to Your AI Tool
The Google Search Console MCP works with any MCP-compatible client. Add the server URL and start asking about organic search in natural language.
Claude Pro / Max
Recommended- 1.Open claude.ai/customize/connectors
- 2.Click "Add custom connector"
- 3.URL:
https://google-search-console.mcp.pipeboard.co/
Claude Code
CLI / VS CodeRun this command in your terminal:
claude mcp add --transport http pipeboard-google-search-console https://google-search-console.mcp.pipeboard.co/Then type /mcp in Claude Code to authenticate with Pipeboard.
Alternative: use an API token for instant setup (no browser auth needed):
https://google-search-console.mcp.pipeboard.co/?token=pipeboard_...Get your token at pipeboard.co/api-tokens
Claude Desktop JSON configuration:
{
"mcpServers": {
"google-search-console": {
"url": "https://google-search-console.mcp.pipeboard.co/"
}
}
}Try these prompts
Losing ground
"Compare the last 28 days to the 28 before it and show me the 20 queries that lost the most clicks"
Almost there
"Find queries in positions 5 to 15 with over 1,000 impressions — which pages should I improve first?"
Index check
"Inspect this URL in Search Console and tell me whether it is indexed, what the canonical is, and when it was last crawled"
Sitemap health
"List my sitemaps, when Google last read each one, and flag any with errors or warnings"
What’s included
Reporting (read-only)
- Search performance by query, page, country, device, date and search appearance
- URL inspection — index status, canonical, crawl date, mobile usability
- Property discovery across every connected Google account
- Sitemap listing and per-sitemap status, counts and errors
Management (opt-in writes)
- Submit a sitemap, or delete one you no longer publish
- Add a property to Search Console, or remove one
- Requires the write scope, which you grant separately
- Never touches your historical performance data
Best Practices
- List your properties first — the exact site_url (sc-domain:example.com or https://www.example.com/) is what every other tool keys on
- Search Console performance data lags by two to three days; end your date ranges accordingly or the last rows will look like a drop
- Prefer domain properties (sc-domain:) when you have them — a URL-prefix property misses traffic on other subdomains and schemes
- Search Console caps a query report at 1,000 rows per request; filter by page or country to get past the ceiling instead of asking for everything
- Queries below a privacy threshold are withheld by Google, so per-query clicks will not add up to the property total — that gap is expected
- Connect every Google account that owns properties; discovery aggregates across all of them so nothing is silently missing
Frequently asked questions
What is a Google Search Console MCP server?
It is a Model Context Protocol server that gives an AI assistant direct access to your Search Console data. Instead of exporting rows from the Performance report, you ask a question — "which queries lost clicks this month?" — and the assistant calls the Search Console API and answers from your real data. The Pipeboard Google Search Console MCP is available at https://google-search-console.mcp.pipeboard.co/ and works with Claude, ChatGPT and any MCP-compatible client.
How is this different from the free open-source GSC MCP servers?
The open-source servers are genuinely capable and worth using if you are comfortable running one. The difference is setup and scale: they run on your machine and ask you to create your own Google Cloud project, enable the Search Console API and manage OAuth credentials yourself. Pipeboard is hosted — you click Connect, approve Google’s consent screen, and the connection works from any client on any device. It also handles several Google accounts at once, restricts which properties a connection may reach, refreshes tokens for you, and puts Search Console in the same session as your Google Ads and Meta Ads data.
What permissions does it ask for?
Read-only by default. A new connection requests webmasters.readonly, which covers every reporting tool — performance queries, index inspection, listing properties and sitemaps. If you want the assistant to submit sitemaps or add and remove properties, you enable write access separately from your connections page, which re-runs Google consent for the broader webmasters scope. Nothing writes to Search Console unless you have explicitly granted that.
Can I use it with more than one Google account?
Yes. Connect as many Google accounts as you need. Property discovery aggregates across all of them and every result is labelled with the account that can reach it, so an agency managing properties under different logins sees one combined list rather than switching accounts.
Does it work with Claude and ChatGPT?
Yes, and with Claude Code, Cursor, n8n, and any client that speaks MCP over Streamable HTTP. Claude and other clients that support dynamic client registration complete OAuth automatically; for everything else you can use a Pipeboard API token in the URL.
How fresh is the data?
It comes straight from the Search Console API on every call, so it is exactly as fresh as Search Console itself — which typically means performance data lags by two to three days. There is no separate sync or warehouse in between, and nothing is cached between your question and Google.