SEO MCP Connector
Local management console and permission wrapper for integrating Google Search Console, GA4, Bing Webmaster Tools, PageSpeed, and SEO analysis through a controlled MCP endpoint.
README
SEO MCP Connector
SEO MCP Connector is a local management console and permission wrapper for search-console-mcp. It lets an MCP client use Google Search Console, Google Analytics 4, Bing Webmaster Tools, PageSpeed, and SEO analysis tools through one controlled local entry point.
The wrapper is designed for agent use: agents connect to bin/run-seo-mcp.js, while the dashboard controls credentials, site allowlists, and which tool categories are exposed.
Features
- Local dashboard for OAuth client details, Bing API keys, service account paths, allowlists, and permission switches.
- Encrypted local secret storage using AES-256-GCM.
- MCP tool filtering at both
tools/listandtools/call. - Defaults to read-only SEO and analytics workflows; write, delete, and submit actions are disabled by default.
- Portable engine installer for
search-console-mcp@1.14.0. - Bilingual user and agent documentation.
Privacy Model
This project is local-first. It does not ship with credentials, tokens, account emails, domains, GA4 property IDs, or API keys.
Ignored local runtime files:
.seo-mcp-keydata/config.jsondata/runtime.jsondata/users.jsonvendor/.env
Do not commit or share those files. They are generated on each machine.
Requirements
- Node.js 20 or newer.
- A Google Cloud OAuth client. A Desktop app client is recommended for local OAuth.
- Enabled Google APIs:
- Google Search Console API
- Google Analytics Data API
- Google Analytics Admin API
- Optional: Bing Webmaster Tools API key.
Install
git clone https://github.com/<owner>/seo-mcp-connector.git
cd seo-mcp-connector
npm run prepare-engine
npm start
Open the local URL printed in the terminal, for example:
SEO MCP Connector is running at http://127.0.0.1:54321
There is no default username or password. The first visit asks you to create a local admin account for this machine.
Connect Google Search Console and GA4
- Open the dashboard.
- Save your Google OAuth client ID and client secret.
- Run Google Search Console OAuth:
npm run setup:gsc
- If you need Search Console write operations such as sitemap submission or site management, reauthorize with the full Search Console scope:
npm run setup:gsc:write
Dashboard permissions alone are not enough for write operations. The Google account must also grant https://www.googleapis.com/auth/webmasters.
- Run GA4 OAuth:
npm run setup:ga4
- Check connected accounts:
npm run accounts:list
The setup commands read encrypted dashboard secrets and pass them to the underlying engine as environment variables. OAuth tokens are stored by search-console-mcp on the local machine.
Optional Bing Setup
Add a Bing API key in the dashboard, then run:
npm run setup:bing
Bing read and write permissions are disabled by default until enabled in the dashboard.
MCP Client Configuration
Use the absolute path on your machine:
{
"mcpServers": {
"seo-search-console": {
"command": "node",
"args": [
"/absolute/path/to/seo-mcp-connector/bin/run-seo-mcp.js"
]
}
}
}
Agents should connect to the wrapper above, not directly to npx search-console-mcp.
Default Permissions
Enabled by default:
- Google Search Console read and SEO analysis
- GA4 read
- Sitemap read
- URL Inspection
- PageSpeed and structured data validation
- Cross-platform GSC + GA4 analysis
Disabled by default:
- Bing read
- Site management, which also requires
npm run setup:gsc:write - Sitemap submission, which also requires
npm run setup:gsc:write - Google/Bing indexing submission
- Bing URL submission
- Delete and remove operations
If a disabled tool is guessed by an agent, bin/run-seo-mcp.js rejects the call.
Documentation
- Chinese guide:
README.zh-CN.md - English agent guide:
docs/AGENT_USAGE.md - Chinese agent guide:
docs/AGENT_USAGE.zh-CN.md
Maintenance
npm run check
npm run prepare-engine
npm run accounts:list
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。