Google Search Console MCP
MCP server for the Google Search Console API. Enables search performance analytics, sitemap management, URL index inspection, and property management through natural language.
README
Google Search Console MCP
MCP server for the Google Search Console API: search performance analytics (clicks, impressions, CTR, position), sitemap management, URL index inspection and property management — from Claude, Cursor, Codex and other AI clients, in natural language.
Ask your assistant "which queries brought the most clicks last month?", "is this page indexed?", or "resubmit the sitemap" — it drives the Search Console API for you, from a quick indexing check to a full performance analysis.
Quick start
-
Get OAuth credentials for the Search Console API.
-
Add the server — for example, in Claude Code (other clients):
claude mcp add google-search-console \ -e GOOGLE_SEARCH_CONSOLE_CLIENT_ID=your_client_id \ -e GOOGLE_SEARCH_CONSOLE_CLIENT_SECRET=your_client_secret \ -e GOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN=your_refresh_token \ -- npx -y mcp-google-search-console@latest -
Ask the assistant: "Show me the top 20 search queries for example.com over the last 28 days, with clicks and CTR."
Tools
| Tool | Description |
|---|---|
list_sites |
All Search Console properties the account can access, with permission levels. |
get_site |
One property's entry (siteUrl + permissionLevel). |
add_site |
Add a property (stays unverified until verified in Search Console). |
delete_site |
Unlink a property from the account (no data is deleted). |
search_analytics |
Performance query: clicks/impressions/CTR/position by date, query, page, country, device, search appearance or hour; filters, search type, pagination, fresh-data mode. |
get_top_queries |
Shortcut for the most common ask: top queries by clicks, optionally filtered by page substring, country or device. |
list_sitemaps |
Submitted sitemaps (or children of a sitemap index), with errors/warnings. |
get_sitemap |
One sitemap's status and per-content-type counts. |
submit_sitemap |
Submit or resubmit a sitemap. |
delete_sitemap |
Remove a sitemap from Search Console. |
inspect_url |
Google-index status of a URL: verdict, coverage, crawl info, canonicals, rich results. |
raw_request |
Escape hatch: any Search Console API path (webmasters/v3 or v1). |
Plus resilience built in: automatic OAuth token refresh (including on 401), retries with backoff on 429 (and on 5xx/network errors for reads only — mutations are never replayed), a request timeout, and an SSRF guard so the token can't leak to a foreign host.
Example prompts
- "Which pages lost the most clicks this month compared to the previous one?"
- "Is https://example.com/pricing indexed? If not, why?"
- "List queries containing 'mcp' where we rank below position 10."
- "Do any of our sitemaps have errors? Resubmit the ones that do."
Two property formats (the #1 pitfall)
The site_url must match the property exactly as registered in Search Console:
- URL-prefix property — full URL with scheme and trailing slash:
https://example.com/. Notehttp://example.com/,https://example.com/andhttps://www.example.com/are three different properties. - Domain property —
sc-domain:example.com(no scheme, no slash).
A near-match returns 403/404. list_sites shows the exact registered values —
the assistant will usually call it first.
Limitations (the API's, not the server's)
- Dates are in Pacific Time and
end_dateis inclusive; final analytics data lags ~2–3 days (passdata_state: "all"for fresh, still-changing rows). - Filters are AND-only — no OR across filters; regex operators use RE2 syntax.
- URL inspection quota is small: 2,000 inspections per property per day (600/minute) — don't batch-inspect entire sites.
- Search analytics: max 25,000 rows per request (paginate with
start_row); anonymized long-tail queries are never returned. - Verification is out of scope:
add_siteregisters a property, but verification happens in the Search Console UI or the Site Verification API.
Installation
Requires Node.js 20+ (runs via npx, no separate install).
<details open> <summary><b>Claude Code</b></summary>
claude mcp add google-search-console \
-e GOOGLE_SEARCH_CONSOLE_CLIENT_ID=your_client_id \
-e GOOGLE_SEARCH_CONSOLE_CLIENT_SECRET=your_client_secret \
-e GOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN=your_refresh_token \
-- npx -y mcp-google-search-console@latest
</details>
<details> <summary><b>Claude Desktop</b></summary>
claude_desktop_config.json — macOS ~/Library/Application Support/Claude/, Windows %APPDATA%\Claude\
{
"mcpServers": {
"google-search-console": {
"command": "npx",
"args": ["-y", "mcp-google-search-console@latest"],
"env": {
"GOOGLE_SEARCH_CONSOLE_CLIENT_ID": "your_client_id",
"GOOGLE_SEARCH_CONSOLE_CLIENT_SECRET": "your_client_secret",
"GOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN": "your_refresh_token"
}
}
}
}
</details>
<details> <summary><b>Cursor</b></summary>
~/.cursor/mcp.json (or .cursor/mcp.json in the project)
{
"mcpServers": {
"google-search-console": {
"command": "npx",
"args": ["-y", "mcp-google-search-console@latest"],
"env": {
"GOOGLE_SEARCH_CONSOLE_CLIENT_ID": "your_client_id",
"GOOGLE_SEARCH_CONSOLE_CLIENT_SECRET": "your_client_secret",
"GOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN": "your_refresh_token"
}
}
}
}
</details>
<details> <summary><b>VS Code</b></summary>
.vscode/mcp.json — note the servers key (not mcpServers)
{
"servers": {
"google-search-console": {
"type": "stdio",
"command": "npx",
"args": ["-y", "mcp-google-search-console@latest"],
"env": {
"GOOGLE_SEARCH_CONSOLE_CLIENT_ID": "your_client_id",
"GOOGLE_SEARCH_CONSOLE_CLIENT_SECRET": "your_client_secret",
"GOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN": "your_refresh_token"
}
}
}
}
</details>
Getting credentials
The Search Console API has no API-key access — every call needs OAuth 2.0 (the data belongs to your account). One-time setup, ~10 minutes:
- Create a Google Cloud project (or reuse one) at console.cloud.google.com, then enable the Google Search Console API: APIs & Services → Library → Google Search Console API → Enable.
- Configure the OAuth consent screen (APIs & Services → OAuth consent screen): choose External, fill in the app name and your email, and add your Google account under Test users (in Testing mode only listed users can authorize — no app verification needed).
- Create an OAuth client (APIs & Services → Credentials → Create credentials → OAuth client ID), application type Desktop app. Save the client ID and client secret.
- Mint a refresh token. The easiest way is the
OAuth 2.0 Playground:
- Click the gear icon → check Use your own OAuth credentials → paste
your client ID and secret (add
https://developers.google.com/oauthplaygroundas an authorized redirect URI to the OAuth client first). - In Step 1, enter the scope
https://www.googleapis.com/auth/webmastersand click Authorize APIs, signing in with the test-user account that owns your Search Console properties. - In Step 2, click Exchange authorization code for tokens and copy the refresh token.
- Click the gear icon → check Use your own OAuth credentials → paste
your client ID and secret (add
- Put the three values into
GOOGLE_SEARCH_CONSOLE_CLIENT_ID,GOOGLE_SEARCH_CONSOLE_CLIENT_SECRETandGOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN. The server exchanges the refresh token for short-lived access tokens automatically.
Scope notes: https://www.googleapis.com/auth/webmasters covers everything
including sitemap submission and property management. If you only need reads
(analytics, inspection, listings), the narrower
https://www.googleapis.com/auth/webmasters.readonly works — but add_site,
delete_site, submit_sitemap and delete_sitemap will then return 403.
While the consent screen stays in Testing mode, refresh tokens expire after
7 days — publish the app (or keep it Internal in a Workspace domain) for
long-lived tokens.
⚠️ The credentials are stored in plain text in your client's config — treat them like a password. The refresh token grants access to your Search Console data until revoked at myaccount.google.com/permissions.
Configuration
| Variable | Required | Default | Description |
|---|---|---|---|
GOOGLE_SEARCH_CONSOLE_CLIENT_ID |
yes* | — | OAuth2 client id (refresh flow). |
GOOGLE_SEARCH_CONSOLE_CLIENT_SECRET |
yes* | — | OAuth2 client secret (refresh flow). |
GOOGLE_SEARCH_CONSOLE_REFRESH_TOKEN |
yes* | — | OAuth2 refresh token (refresh flow). |
GOOGLE_SEARCH_CONSOLE_ACCESS_TOKEN |
yes* | — | Alternative: a static access token (~1 h lifetime), mostly for testing. |
GOOGLE_SEARCH_CONSOLE_API_BASE |
no | https://searchconsole.googleapis.com |
API root override. |
GOOGLE_SEARCH_CONSOLE_TIMEOUT_MS |
no | 60000 |
Per-request timeout, ms. |
GOOGLE_SEARCH_CONSOLE_MAX_RETRIES |
no | 3 |
Retries on transient errors. |
* Either the three refresh-flow variables together, or GOOGLE_SEARCH_CONSOLE_ACCESS_TOKEN.
Documentation
- All tools — full reference with parameters and notes.
- Development — build, tests, smoke check, telemetry.
- Publishing — releasing and MCP-catalog listing.
Support
Questions, ideas, issues — Telegram @gistrec or GitHub issues.
License
MIT — see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。