CloakBrowser Search MCP
A local web search MCP server that uses stealth Chromium to bypass CAPTCHAs and bot blocks, supporting Google and DuckDuckGo with automatic fallback.
README
CloakBrowser Search MCP
A 100% local, anti-bot-resilient web search Model Context Protocol (MCP) server utilizing cloakbrowser (a stealth Chromium automation tool).
Features
- Dual Search Engines: Search Google (default) or DuckDuckGo.
- Stealth Interaction: Google searches simulate natural human behavior (typing speed, keyboard navigation, and viewport configurations) to completely bypass CAPTCHAs and bot blocks.
- Self-Healing Fallback: Automatically switches to DuckDuckGo if Google blocks requests, ensuring 100% reliable results.
- Zero API Keys: Free, local, and private scraping.
Quick Start (No Installation Needed)
You can run this server directly via npx in your MCP client configurations.
(On first execution, cloakbrowser will automatically download its stealth Chromium binary ~535 MB in the background).
1. Cursor Setup
-
Register the MCP server in Settings -> Features -> MCP:
- Name:
cloak-search-mcp - Type:
command - Command:
npx -y cloak-search-mcp
- Name:
-
To route Cursor to your local LLM:
- Go to Settings -> Models.
- Under OpenAI API Key, enter a dummy key (e.g.,
sk-no-key-required). - Click Configure and set Override Base URL to:
- Llama.cpp:
http://localhost:8080/v1 - Ollama:
http://localhost:11434/v1
- Llama.cpp:
- Add your local model name (e.g.,
default_modelorllama3) and disable Cursor's default cloud models.
2. Copilot CLI Setup
-
Set environment variables to route Copilot CLI to your local LLM:
- PowerShell (Llama.cpp):
$env:COPILOT_PROVIDER_BASE_URL="http://127.0.0.1:8080" $env:COPILOT_OFFLINE="true" $env:COPILOT_MODEL="default_model" - PowerShell (Ollama):
$env:COPILOT_PROVIDER_BASE_URL="http://127.0.0.1:11434/v1" $env:COPILOT_OFFLINE="true" $env:COPILOT_MODEL="llama3" # Replace with your loaded Ollama model - Bash (Llama.cpp):
export COPILOT_PROVIDER_BASE_URL="http://127.0.0.1:8080" export COPILOT_OFFLINE="true" export COPILOT_MODEL="default_model" - Bash (Ollama):
export COPILOT_PROVIDER_BASE_URL="http://127.0.0.1:11434/v1" export COPILOT_OFFLINE="true" export COPILOT_MODEL="llama3" # Replace with your loaded Ollama model
- PowerShell (Llama.cpp):
-
Register the MCP server with Copilot CLI:
copilot mcp add cloak-search-mcp -- npx -y cloak-search-mcp
3. Claude CLI Setup
If running a local Claude CLI session against your local model, set the following environment variables:
- PowerShell (Llama.cpp):
$env:ANTHROPIC_BASE_URL="http://127.0.0.1:8080" $env:ANTHROPIC_AUTH_TOKEN="sk-no-key-required" $env:ANTHROPIC_MODEL="default_model" - PowerShell (Ollama):
$env:ANTHROPIC_BASE_URL="http://127.0.0.1:11434/v1" $env:ANTHROPIC_AUTH_TOKEN="sk-no-key-required" $env:ANTHROPIC_MODEL="llama3" # Replace with your loaded Ollama model - Bash (Llama.cpp):
export ANTHROPIC_BASE_URL="http://127.0.0.1:8080" export ANTHROPIC_AUTH_TOKEN="sk-no-key-required" export ANTHROPIC_MODEL="default_model" - Bash (Ollama):
export ANTHROPIC_BASE_URL="http://127.0.0.1:11434/v1" export ANTHROPIC_AUTH_TOKEN="sk-no-key-required" export ANTHROPIC_MODEL="llama3" # Replace with your loaded Ollama model
Global Installation (Alternative)
If you prefer to install it globally on your machine:
npm install -g cloak-search-mcp
Then configure your MCP client to run:
- Command:
cloak-search-mcp
Local Development Setup
If you want to modify or run the code locally:
- Clone the repository.
- Install dependencies:
npm install - Run the server locally:
node index.js
License
This project is licensed under the MIT License - see the LICENSE file for details.
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。