Google Search Console contains some of the best first-party SEO data available: search queries, clicks, impressions, average position, click-through rate, indexing status and sitemap information. A Google Search Console MCP server makes that data available to an AI agent through the Model Context Protocol (MCP).
Instead of downloading CSV files or building scripts against the Google Search Console API, you can ask an MCP-compatible tool to find declining pages, weak click-through rates, striking-distance keywords and indexing problems. A coding agent can then inspect the website project and help implement approved improvements.
This guide explains what a Google Search Console MCP is, how it differs from the API and a traditional CLI, and how to connect SearchConsole.ai to Claude Code, Codex, Cursor or another MCP client.
Quick answer: Google provides a Search Console API, but its official documentation does not currently list a Google-hosted Search Console MCP endpoint. SearchConsole.ai provides a hosted, read-only MCP connection that uses the official API and Google OAuth.
What is a Google Search Console MCP server?
MCP is a standard that lets an AI application discover and call tools provided by an external server. A Google Search Console MCP server translates an agent's tool calls into authenticated Search Console API requests and returns structured results the agent can analyze.
The flow looks like this:
- You authorize access with your Google account.
- Your MCP client connects to the remote server with OAuth.
- The client discovers available Search Console tools.
- The agent requests only the data needed for your task.
- The agent analyzes the results and explains its recommendations.
SearchConsole.ai exposes a read-only tool set. It can retrieve Search Analytics, properties, sitemap details and URL inspection data, but it does not modify your Search Console account.
Search Console MCP vs API vs CLI
| Option | Best for | Setup | How you use it |
|---|---|---|---|
| Google Search Console MCP | AI-assisted SEO analysis and agent workflows | Connect OAuth once | Ask questions in natural language |
| Google Search Console API | Custom software, dashboards and data pipelines | Build an OAuth client and API integration | Send HTTP requests or use a client library |
| Google Search Console CLI | Scripts and terminal automation | Build or install a CLI wrapper | Run commands with flags and parameters |
| Search Console web interface | Manual reporting and investigation | Sign in to Google | Use reports, filters and exports |
The Search Console API is the underlying data source. MCP adds tool definitions and context that help an agent decide which API operation to use. A CLI is useful for deterministic scripts, while MCP is better for exploratory questions and multi-step analysis.
Is there a free Google Search Console MCP?
Search Console itself and its API do not require a separate paid API subscription, although Google applies usage quotas. SearchConsole.ai currently provides a hosted connection without requiring users to create an API key or deploy a server.
A self-hosted Google Search Console MCP from GitHub can also be inexpensive, but you must operate the server, create and secure Google OAuth credentials, store refresh tokens, handle updates and configure every client. A hosted service removes most of that setup.
How to connect Google Search Console to AI tools
1. Connect your Google account
- Open SearchConsole.ai.
- Select Sign in with Google.
- Choose the Google account that has access to the Search Console properties you need.
- Approve read-only access to Search Console.
- Open the account page to see the setup instructions for each supported client.
Google identity and Search Console access are authorized separately. SearchConsole.ai stores the resulting Google token encrypted and uses it only when an authenticated MCP client requests your data.
2. Connect Claude Code
Run these commands in your terminal:
claude mcp add --transport http searchconsole-ai --scope user https://searchconsole.ai/mcp
claude mcp login searchconsole-ai
Approve the SearchConsole.ai connection in the browser. Then test the connection:
Use SearchConsole.ai to list my Google Search Console properties.
This is the simplest Google Search Console MCP Claude Code setup because the connection is registered at user scope and can be reused across projects.
3. Connect Codex
Run:
codex mcp add searchconsole-ai --url https://searchconsole.ai/mcp
Codex should detect OAuth support and open the browser automatically. Complete OAuth, restart Codex if necessary and open the website repository you want to improve.
If Codex does not open the browser or authentication is interrupted, start OAuth explicitly:
codex mcp login searchconsole-ai
You can also add the server from the Codex app under Settings > MCP servers. Choose Streamable HTTP, use searchconsole-ai as the name and enter:
https://searchconsole.ai/mcp
4. Connect Cursor
Install the public SearchConsole.ai Cursor plugin, or add the following entry inside the mcpServers object in ~/.cursor/mcp.json:
{
"mcpServers": {
"searchconsole-ai": {
"type": "http",
"url": "https://searchconsole.ai/mcp"
}
}
}
Open Cursor Settings > Tools & MCP, enable SearchConsole.ai and complete authentication in the browser.
5. Connect another MCP-compatible agent
Use these connection details:
| Setting | Value |
|---|---|
| Name | searchconsole-ai |
| Server URL | https://searchconsole.ai/mcp |
| Transport | Streamable HTTP |
| Authentication | OAuth |
If the agent can configure MCP servers itself, give it this prompt:
Add a user-level remote Streamable HTTP MCP server named searchconsole-ai with URL https://searchconsole.ai/mcp. Authenticate it with OAuth in my browser, then use SearchConsole.ai to list all Google Search Console properties I can access.
What can an AI agent do with Search Console data?
Connecting the server is only the first step. The most useful workflows combine Search Console evidence with the website project the agent can inspect.
Find high-impression, low-click opportunities
Ask the agent to sort queries and pages by impressions, then compare clicks, CTR and average position. High impressions with few clicks can indicate:
- A title or search snippet that does not match intent
- A page ranking below the strongest click positions
- A broad query for which the current page is only partially relevant
- Competing pages splitting relevance
Do not treat every low-CTR query as a title problem. CTR varies by position, device, query intent and SERP features.
Find striking-distance keywords
Queries in positions 4 to 20 can expose pages that already have relevance but need improvement. Ask the agent to group opportunities by page and determine whether each page needs:
- Better intent alignment
- Clearer headings and supporting sections
- Stronger internal links
- Consolidation with overlapping content
- A genuinely new page for a distinct topic
Diagnose traffic declines
Compare equivalent periods and segment the change by query, page, device and country. The agent should distinguish between:
- Lost rankings
- Lower search demand
- Reduced CTR
- Indexing or canonical problems
- Changes limited to one template or section
Review indexing and sitemaps
Use URL inspection and sitemap tools to identify pages that are excluded, canonicalized unexpectedly or missing from submitted sitemaps. Search Console data should guide the investigation, while the website code and rendered output provide the implementation evidence.
A useful first SEO prompt
Start a new chat with the website project for a property available in your Google Search Console account, then use:
Use SearchConsole.ai to audit my SEO and show my highest-impact opportunities. Read and follow the SearchConsole.ai SEO analysis playbook before analyzing the data.
A good answer should show the evidence behind each recommendation, estimate likely impact, separate findings from assumptions and suggest a clear next action. When the agent has access to the website repository, ask it to inspect the relevant templates and propose a reviewable implementation rather than changing pages blindly.
Google Search Console MCP on GitHub
GitHub projects are useful when you need a self-hosted server or want to inspect an implementation. SearchConsole.ai's public Cursor plugin repository contains the Cursor manifest and SEO analysis skill for its hosted MCP endpoint.
Before running any Google Search Console MCP from GitHub, review:
- Which Google OAuth scopes it requests
- Where access and refresh tokens are stored
- Whether its public tools are read-only
- Whether credentials can appear in logs
- How updates and security fixes are delivered
- Whether the repository is actively maintained
Never commit a downloaded Google OAuth client-secret JSON file or user refresh token to Git.
Google Search Console MCP vs Google Analytics MCP
Search Console and Google Analytics answer different questions:
| Data source | Primary questions |
|---|---|
| Google Search Console | Which Google queries and pages earn impressions, clicks and rankings? Is a URL indexed? Are sitemaps processing correctly? |
| Google Analytics | What do visitors do after arriving? Which channels, events and conversions contribute to business outcomes? |
Google provides an official Google Analytics MCP server, including a managed Google Analytics Data API endpoint at https://analyticsdata.googleapis.com/mcp/v1. Google also maintains an experimental Google Analytics MCP project on GitHub.
For a complete SEO analysis, use Search Console for organic search visibility and Analytics for on-site behavior and conversion outcomes. Do not combine the metrics without checking that dates, time zones, attribution and page URLs align.
Frequently asked questions
Does Google Search Console have an MCP?
Google provides the official Search Console API, but its documentation does not currently list a Google-hosted Search Console MCP endpoint. Third-party and hosted servers such as SearchConsole.ai connect MCP clients to the official Search Console API. This status was last reviewed in July 2026.
Does Google support MCP?
Yes. Google supports MCP and provides managed remote MCP servers for a growing range of Google and Google Cloud services. Google announced official MCP support in December 2025 and later expanded the catalog to more than 50 managed servers. Availability and launch status vary by product.
Is there an MCP for Google Workspace?
Yes. Google documents remote MCP servers for Gmail, Google Drive, Google Calendar, Google Chat and the People API. At the time of review, the Google Workspace MCP servers were available through the Google Workspace Developer Preview Program.
Is there an MCP server for Google Analytics?
Yes. Google offers an official read-only Google Analytics MCP server and a managed Analytics Data API MCP endpoint. It can run reports, retrieve metadata, query realtime data and check metric and dimension compatibility.
Can I use Google Search Console with Claude?
Yes. Claude Code can connect to a remote Search Console MCP server over HTTP and authorize through OAuth. After connecting SearchConsole.ai, Claude can list properties, query search performance, inspect URLs and review sitemap information according to your Google permissions.
Is Google Search Console MCP the same as Google Console MCP?
No. "Google Console MCP" is ambiguous and may refer to Google Cloud Console, Search Console or another Google product. Check the server's endpoint, tool list and OAuth scopes before authorizing it. A Search Console MCP should request only the Search Console permissions needed for its advertised tools.
Do I need to create a Google Cloud project?
Not when using a hosted service such as SearchConsole.ai. The service operates its own verified OAuth application. A self-hosted GitHub server usually requires you to create a Google Cloud project, enable the Search Console API and configure OAuth credentials.
Is SearchConsole.ai read-only?
Yes. Its public MCP server advertises read-only Search Console tools. It does not expose tools that add or remove properties, submit or delete sitemaps, or change Google account settings.
Next step
Connect Google Search Console to your AI tools and start by listing your available properties. Then open the corresponding website project in Claude Code, Codex, Cursor or another agent and run the SEO audit prompt above.