crx-mcp

crx-mcp

MCP server for Chrome extension testing. Access chrome.storage, Service Workers, declarativeNetRequest, and more directly from AI coding agents like Claude Code.

Category
访问服务器

README

crx-mcp

MCP server for Chrome extension testing. Access chrome.storage, Service Workers, declarativeNetRequest, and more directly from AI coding agents like Claude Code.

Why?

Existing browser MCP servers (like Playwright MCP) can't access Chrome extension internals. crx-mcp fills this gap with 15 purpose-built tools for extension development and testing.

Quick Start

With Claude Code

Add to your .mcp.json:

{
  "mcpServers": {
    "crx-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "crx-mcp"]
    }
  }
}

Then in Claude Code:

> Load my extension from ./my-extension and check if storage is working

Claude will use extension_load, then storage_get to inspect chrome.storage

CLI

npx crx-mcp --extension-path ./my-extension --verbose

Tools (15)

Core

Tool Description
extension_load Load an unpacked extension and launch Chrome. Returns extension ID.
navigate Navigate to a URL. Returns title and status code.
snapshot Accessibility tree snapshot of page, popup, or side panel.
storage_get Read from chrome.storage (local/sync/session).
storage_set Write to chrome.storage (local/sync/session).
eval_service_worker Execute JavaScript in the Service Worker context.
console_logs Get console logs from all contexts (page, SW, popup, sidepanel).

Extension-Specific

Tool Description
manifest_validate Validate manifest.json against MV3 requirements (no browser needed).
open_popup Open the extension popup and return accessibility snapshot.
open_sidepanel Open the side panel and return accessibility snapshot.
dnr_rules List declarativeNetRequest rules (dynamic/session/static).
permissions_check Compare declared vs granted permissions.

Advanced

Tool Description
screenshot Take a PNG screenshot (base64).
network_requests List captured network requests with URL filter.
content_script_eval Execute JS in page context (ISOLATED or MAIN world).
reload_extension Hot-reload the extension and re-attach to Service Worker.

CLI Options

crx-mcp [options]
  --extension-path <path>   Pre-load extension at startup
  --chrome-path <path>      Path to Chrome executable
  --user-data-dir <path>    Chrome user data directory
  --no-sandbox              Disable sandbox (for CI/Docker)
  --verbose                 Debug logging to stderr

Requirements

  • Node.js >= 18
  • Google Chrome installed (detected automatically on Windows, macOS, Linux)
  • Extensions require headed mode (no headless)

How It Works

  1. extension_load launches Chrome with --load-extension and detects the extension ID
  2. Service Worker communication uses Chrome DevTools Protocol (CDP) sessions
  3. chrome.storage, declarativeNetRequest, and permissions are accessed via JS evaluation in the SW context
  4. Console logs are collected from all contexts (page, SW, popup, sidepanel) into a ring buffer
  5. Network requests are monitored via CDP Network.enable
  6. Popup and side panel are opened in new tabs (actual popup/sidepanel UI requires user gestures)

Use Cases

  • Automated testing of Chrome extensions during development
  • AI-assisted debugging — let Claude inspect storage, logs, and network traffic
  • CI/CD validation — validate manifests and permissions without manual testing
  • Extension factory workflows — quickly test multiple extensions in sequence

License

MIT

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选