A free connector that lets Claude, ChatGPT and Cursor read your Google Search Console data. Every property your Google account can see, the full 16 months, no Google Cloud project to set up. It is read-only, and the code is open source.
https://gsc.modelbrain.net/mcpMore free connectors will be listed on this page as they launch.
You can name a site the way you would say it, such as example.com. The connector matches it to the right Search Console property.
| Tool | What it returns |
|---|---|
| list_sites | Every property the Google account can see, with its permission level. |
| search_performance | Clicks, impressions, CTR and position by search, page, country, device, date or search appearance, with filters. Ranked by clicks, then impressions. |
| compare_periods | Two date ranges side by side, with the searches or pages that gained and lost the most. |
| inspect_url | Index status for one URL: coverage, last crawl, Google's chosen canonical, robots.txt and rich results. |
| list_sitemaps | Submitted sitemaps, when Google last read them, errors, warnings and URL counts. |
https://gsc.modelbrain.net/mcp.https://gsc.modelbrain.net/mcp and sign in with Google.ChatGPT's menu names change between plans and releases, so the labels in your account may differ.
Add it to your mcp.json, then sign in when Cursor asks:
{
"mcpServers": {
"modelbrain-search-console": {
"url": "https://gsc.modelbrain.net/mcp"
}
}
}
On Google's permission screen, tick "View Search Console data for your verified sites". Google leaves that box unticked, and without it the connector cannot read anything.
Your Google account identifier, the access token Google issues (encrypted), the name and return address of each app you connect, and its own sign-in tokens, stored as hashes. Your Search Console data goes straight to the app that asked for it and is never saved. A connection unused for 90 days is deleted. The full terms are in section 7 of the privacy policy, and the code is on GitHub under the MIT licence.
Keep what you learn from it
The connector reads today's numbers. ModelBrain is the memory layer that keeps what you and your assistant concluded about them, on your own computer, so next month you can ask what changed since you last looked, in Claude, ChatGPT or Cursor.
Download ModelBrain (beta)Yes, for every property your Google account can see, with the full 16 months Search Console keeps. There is no paid tier. Each account can make up to 60 tool calls a minute, and Google allows up to 25,000 rows in one report.
No. It asks Google only for read-only access, so it cannot submit sitemaps, remove URLs or change settings. Google does not offer Request Indexing to apps for ordinary pages, so that stays in Search Console itself.
No. Results go straight to the app that asked for them. The connector keeps your Google account identifier, an encrypted access token and its own sign-in tokens for the apps you connect. Section 7 of the privacy policy has the details.
Remove the connector in your app, and remove ModelBrain Connectors at myaccount.google.com/permissions. Once Google refuses the access, the connector deletes your record.
Yes. The connector is open source under the MIT licence at github.com/ModelBrain/modelbrain-connectors.
Something not working? Write to hello@modelbrain.net.