Google Search Console Mcp

Search Console MCP Server -- Google Search Analytics for AI Assistants

[](https://github.com/ncosentino/google-search-console-mcp/releases/latest) [](LICENSE) [](go/go.mod) [](csharp/Directory.Build.props) [](https://github.com/ncosentino/google-search-console-mcp/actions/workflows/ci.yml)

> Zero-dependency MCP server for Google Search Console. > Pre-built native binaries for Linux, macOS, and Windows. No Node.js. No Python. No .NET runtime. No Go toolchain. Download one binary and configure your AI tool.

Expose real Google organic search data directly to AI assistants like Claude, GitHub Copilot, and Cursor via the Model Context Protocol (MCP). Ask your AI which queries are driving traffic to which pages, identify ranking opportunities, and diagnose CTR issues -- all grounded in real Search Console data for your specific property.

---

Why This Exists

AI assistants are powerful at analyzing SEO strategy -- but they need real data. This MCP server bridges your AI tool to Google Search Console, giving it:

• Real organic search performance -- clicks, impressions, CTR, and average position for every query and page in your site
• Up to 50,000 rows per query (vs 1,000 in the UI), unlocking complete keyword datasets
• 16 months of history for trend analysis
• Dimension flexibility -- group by query, page, country, device, date, or any combination

With this MCP server configured, you can ask your AI: "Which queries am I ranking position 8-15 for with high impressions but low CTR? What pages could I optimize to break into the top 5?" and get a real data-backed answer.

---

Quick Start

Three steps: create a service account, download a binary, add it to your MCP config.

Step 1: Create a Google Service Account

1. Go to Google Cloud Console and create or select a project
2. Enable the Google Search Console API for that project:

• Go to https://console.cloud.google.com/apis/library/searchconsole.googleapis.com
• Click Enable

1. Create a Service Account (IAM & Admin → Service Accounts → Create Service Account)

• Give it a name (e.g. gsc-mcp) -- no project-level roles needed

1. Click on the service account → Keys tab → Add Key → Create new key → JSON

• Download the JSON file -- the client_email field inside is the email you'll use in step 5

1. Go to Google Search Console → select your property → Settings → Users and permissions → Add user

• Enter the client_email from the JSON file exactly (e.g. gsc-mcp@my-project.iam.gserviceaccount.com)
• Set permission to Full → Add

> Note: When adding the service account in Search Console, use the exact email from the client_email field in the downloaded JSON -- not a manually constructed one.

> Note on property URL format: Search Console has two property types. If your property was added as a domain property, the site URL is sc-domain:example.com (not https://www.example.com/). Use list_sites to discover the correct format for your property.

> The Search Console API is free. No billing account is required.

Step 2: Download a Binary

Go to the Releases page and download the binary for your platform:

| Platform | Go binary | C# binary | |----------|-----------|-----------| | Linux x64 | gsc-mcp-go-linux-amd64 | gsc-mcp-csharp-linux-x64 | | Linux arm64 | gsc-mcp-go-linux-arm64 | gsc-mcp-csharp-linux-arm64 | | macOS x64 (Intel) | gsc-mcp-go-darwin-amd64 | gsc-mcp-csharp-osx-x64 | | macOS arm64 (Apple Silicon) | gsc-mcp-go-darwin-arm64 | gsc-mcp-csharp-osx-arm64 | | Windows x64 | gsc-mcp-go-windows-amd64.exe | gsc-mcp-csharp-win-x64.exe | | Windows arm64 | gsc-mcp-go-windows-arm64.exe | gsc-mcp-csharp-win-arm64.exe |

On Linux/macOS, make the binary executable after downloading:

chmod +x gsc-mcp-go-linux-amd64

Step 3: Add to Your AI Tool Config

See the [Setup by Tool](#setup-by-tool) section below for your specific client.

---

Setup by Tool

Replace /path/to/binary with the actual path to your downloaded binary. Replace /path/to/service-account.json with the path to your downloaded service account JSON key file.

Claude Code / GitHub Copilot CLI

{
  "mcpServers": {
    "search-console": {
      "type": "stdio",
      "command": "/path/to/gsc-mcp-go-linux-amd64",
      "args": [],
      "env": {
        "GOOGLE_SERVICE_ACCOUNT_FILE": "/path/to/service-account.json"
      }
    }
  }
}

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "search-console": {
      "command": "/path/to/gsc-mcp-go-darwin-arm64",
      "env": {
        "GOOGLE_SERVICE_ACCOUNT_FILE": "/path/to/service-account.json"
      }
    }
  }
}

Cursor

{
  "mcpServers": {
    "search-console": {
      "command": "/path/to/gsc-mcp-go-linux-amd64",
      "env": {
        "GOOGLE_SERVICE_ACCOUNT_FILE": "/path/to/service-account.json"
      }
    }
  }
}

VS Code with GitHub Copilot

{
  "mcp": {
    "servers": {
      "search-console": {
        "type": "stdio",
        "command": "/path/to/gsc-mcp-go-linux-amd64",
        "env": {
          "GOOGLE_SERVICE_ACCOUNT_FILE": "/path/to/service-account.json"
        }
      }
    }
  }
}

Visual Studio

{
  "search-console": {
    "command": "C:\\path\\to\\gsc-mcp-csharp-win-x64.exe",
    "env": {
      "GOOGLE_SERVICE_ACCOUNT_FILE": "C:\\path\\to\\service-account.json"
    }
  }
}

Using CLI Argument

You can also pass the service account file path as a CLI argument:

{
  "command": "/path/to/binary",
  "args": ["--service-account-file", "/path/to/service-account.json"]
}

---

Available Tools

query_search_analytics

Query clicks, impressions, CTR, and average position from Search Console.

Parameters:

| Name | Type | Required | Default | Description | |------|------|----------|---------|-------------| | site_url | string | Yes | -- | The Search Console property URL (e.g. https://www.example.com/ or sc-domain:example.com) | | start_date | string | Yes | -- | Start date in YYYY-MM-DD format | | end_date | string | Yes | -- | End date in YYYY-MM-DD format | | dimensions | string[] | No | [] | Group by: query, page, country, device, date. Empty array returns aggregate totals. | | row_limit | int | No | 1000 | Maximum rows to return (1-25000) |

Example prompts:

> "Which queries are driving the most impressions to my site this month but have a CTR below 2%?"

> "Show me my top 50 pages by clicks over the last 90 days."

> "Which queries am I ranking position 8-15 for? These are my best ranking improvement opportunities."

list_sites

List all Search Console properties the service account has access to.

No parameters required.

Example prompt:

> "What Search Console properties do I have access to?"

list_sitemaps

List submitted sitemaps for a property and their status.

Parameters:

| Name | Type | Required | Description | |------|------|----------|-------------| | site_url | string | Yes | The Search Console property URL |

Example prompt:

> "Show me all submitted sitemaps for my site and whether any have errors."

Response Structure

query_search_analytics returns:

{
  "siteUrl": "https://www.example.com/",
  "startDate": "2026-01-01",
  "endDate": "2026-01-31",
  "dimensions": ["query", "page"],
  "rowCount": 1234,
  "rows": [
    {
      "keys": ["blazor dependency injection", "https://www.example.com/blazor-di"],
      "clicks": 142,
      "impressions": 3820,
      "ctr": 0.0372,
      "position": 4.3
    }
  ],
  "queriedAt": "2026-02-21T19:00:00Z"
}

---

Configuration Reference

Credential resolution uses this priority order (highest to lowest):

1. CLI Argument (Highest Priority)

/path/to/gsc-mcp-go-linux-amd64 --service-account-file /path/to/key.json

2. Environment Variable (File Path)

export GOOGLE_SERVICE_ACCOUNT_FILE=/path/to/service-account.json

3. Environment Variable (JSON Content)

export GOOGLE_SERVICE_ACCOUNT_JSON='{"type":"service_account","client_email":"..."}'

Useful in environments where you store secrets as string values rather than files (e.g. containers, CI/CD).

4. .env File (Lowest Priority -- Dev Convenience)

Create a .env file in the working directory:

GOOGLE_SERVICE_ACCOUNT_FILE=/path/to/service-account.json

Or with inline JSON:

GOOGLE_SERVICE_ACCOUNT_JSON={"type":"service_account",...}

---

Transports

Both binaries default to stdio (everything above uses it) and also support --transport http for remote/networked deployments:

./gsc-mcp-go-linux-amd64 --transport http --service-account-file /path/to/key.json

The server listens on the PORT environment variable (default 8080). Point an HTTP-capable MCP client at the server instead of launching it as a subprocess:

{
  "mcpServers": {
    "search-console": {
      "type": "http",
      "url": "http://localhost:8080/"
    }
  }
}

Security defaults: neither Go's net/http nor C#'s Kestrel validate the Host header by default, which would otherwise allow DNS rebinding against a locally-bound server. Both binaries reject any Host header outside localhost, 127.0.0.1, [::1] unless you explicitly widen the allow-list -- --allowed-hosts (comma-separated) for Go, the standard AllowedHosts ASP.NET Core config key (semicolon-separated, settable via --AllowedHosts on the command line) for C#. Go additionally rejects genuinely cross-site browser requests (CSRF) via http.CrossOriginProtection, while allowing same-origin and non-browser (no Origin header) traffic; C# has no equivalent yet.

This is a transport flag, not a hosting product -- no Dockerfile, no cloud-provider automation, and no authentication in front of the MCP endpoint beyond the protections above. TLS termination and network exposure are your responsibility if you deploy either binary over HTTP. See the Transports doc for the full reference.

---

Go vs C# -- Which Binary?

Both implementations expose identical tools with identical behavior.

| Aspect | Go | C# Native AOT | |--------|----|----| | Binary size | ~10-15 MB | ~25-40 MB | | Startup time | ~10-50ms | ~50-100ms | | Runtime dependency | None | None | | Language | Go 1.26 | C# / .NET 10 | | MCP SDK | Official go-sdk | Official ModelContextProtocol | | Auth | golang.org/x/oauth2/google | Native RSA + HttpClient | | Transports | stdio, HTTP | stdio, HTTP |

Recommendation: Both work great. Pick Go for smaller binary size and faster startup; pick C# if you prefer the .NET ecosystem. Both support the same [HTTP transport](#transports).

---

Building from Source

Go

Requires Go 1.26+:

cd go
go mod tidy
go build -ldflags="-s -w" -trimpath -o gsc-mcp-go .

Run tests:

go test ./...

C# (.NET 10 SDK Required)

cd csharp

# Build (non-AOT, for development)
dotnet build SearchConsoleMcp.slnx

# Publish Native AOT
dotnet publish src/SearchConsoleMcp/SearchConsoleMcp.csproj -r linux-x64 -c Release --self-contained true

# Run tests
dotnet test SearchConsoleMcp.slnx

---

Related Projects

• google-psi-mcp -- Zero-dependency MCP server for Google PageSpeed Insights Core Web Vitals
• google-keyword-planner-mcp -- Zero-dependency MCP server for Google Ads Keyword Planner (keyword ideas, search volume, CPC)

---

About

Nick Cosentino -- Dev Leader

This MCP server was built by Nick Cosentino, a software engineer and content creator known as Dev Leader. Nick creates practical .NET, C#, ASP.NET Core, Blazor, and software engineering content for intermediate to advanced developers -- covering everything from performance optimization and clean architecture to real-world career advice.

This tool was born out of real work analyzing search performance for devleader.ca and the desire to use AI assistants effectively during that process. It serves as a practical example of building Native AOT C# and idiomatic Go MCP servers with zero runtime dependencies.

Find Nick online:

• Blog: https://www.devleader.ca
• YouTube: https://www.youtube.com/@devleaderca
• Newsletter: https://weekly.devleader.ca
• LinkedIn: https://linkedin.com/in/nickcosentino
• All My Links: https://links.devleader.ca

BrandGhost

BrandGhost is a social media automation platform built by Nick that lets content creators cross-post and schedule content across all social platforms in one click. If you create content and want to spend less time on distribution and more time creating, check it out.

---

Contributing

Contributions are welcome! Please:

1. Open an issue describing the bug or feature request before submitting a PR
2. Run golangci-lint run (Go) or dotnet build with zero warnings (C#) before submitting
3. Keep both implementations in sync -- a feature added to Go should also be added to C#, and vice versa

---

License

MIT License -- see [LICENSE](LICENSE) for details.

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: ncosentino
• Source: ncosentino/google-search-console-mcp
• License: MIT

Install and usage instructions live in the source repository linked above.