Search

一个基于MCP协议的搜索服务实现，提供网络搜索和本地搜索功能，Cursor和Claude Desktop能与之无缝集成。

python

77

GitHub Stars

python

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "fengin-search-server": {
      "command": "uv",
      "args": [
        "--directory",
        "D:\\code\\search-server",
        "run",
        "search"
      ],
      "env": {
        "BRAVE_API_KEY": "YOUR_BRAVE_API_KEY"
      }
    }
  }
}

You run a modular MCP server that exposes multiple search engines through a single interface. It enables Cursor and Claude Desktop to fetch up-to-date web content by routing requests to Brave Search, Metaso, or Bocha as needed, while keeping engines isolated for easy reuse or replacement.

How to use

You operate this MCP server behind an MCP client. With the server running, your client can request searches, location searches, or scholarly searches depending on the enabled engine. The server consolidates results from the chosen engine and returns structured data to your client, while keeping engines modular so you can reuse or swap individual components without affecting the rest.

How to install

Prerequisites and environment requirements are in place so you can run the MCP server smoothly.

# Prerequisites
Python 3.10+
uv 0.24.0+
node.js v20.15.0
cursor >=0.45.10

# Optional (Metaso requires browser driver)
pip install playwright>=1.35.0
playwright install chromium

# Install the MCP server code (example repository)
# (Clone the project containing the MCP server)
# git clone https://github.com/fengin/search-server.git

Additional sections

Configure and run the server to enable a specific search engine. The runtime supports enabling a single engine at a time for a given deployment. You configure the active engine by setting an environment variable or editing the configuration to select Brave, Metaso, or Bocha.

Configuration and runtime specifics

The server includes dedicated modules for Brave, Metaso, and Bocha, each with its own configuration file. The recommended runtime integration for Claudе Desktop and Cursor shows how to start and pass environment variables. Use the following MCP server configuration to expose the search endpoint to your MCP clients.

Troubleshooting

If your MCP client reports a red status or Tools Not Found after configuring the MCP Server, verify the following:

The required software and versions match the environment section
The command path and working directory are correct when starting the server
The MCP command or URL you reference is exactly as shown in the configuration examples
If you updated configuration files, restart the MCP client to apply changes
If Chromium is not installed, install it as described in the environment setup section.

Available tools

brave_search

Core interface to Brave Search enabling web and location searches with pagination and filters.

metaso_search

Core interface to Metaso Search providing web and scholarly search modes including concise, detail, and (future) research modes.

bocha_search

Core interface to Bocha Search offering web search with time-filtered results, summaries, and image results.