MCP Ferris Search

A blazing-fast MCP (Model Context Protocol) server for multi-engine web search, written in Python.

This is a Python implementation inspired by the original ferris-search Rust project.

Features

• Multi-engine fan-out — search across multiple engines simultaneously with a single call
• 9 search engines — Bing, DuckDuckGo, Brave, Baidu, CSDN, Juejin, Zhihu, GitHub, GitHub Code
• 7 MCP tools — web_search + 6 content fetchers
• No API keys required for most engines (Brave requires API key)
• Proxy support — HTTP/SOCKS5 proxy via environment variables

Installation

From PyPI (recommended)

pip install mcp-ferris-search

Using uvx

uvx mcp-ferris-search

From source

git clone https://github.com/your-username/mcp-ferris-search.git
cd mcp-ferris-search
pip install -e .

Configuration

Claude Desktop / Cursor

Add to your MCP settings:

{
  "mcpServers": {
    "ferris-search": {
      "command": "uvx",
      "args": ["mcp-ferris-search"],
      "env": {
        "DEFAULT_SEARCH_ENGINE": "bing"
      }
    }
  }
}

Environment Variables

Variable	Default	Description
DEFAULT_SEARCH_ENGINE	bing	Default engine when engines param is omitted
ALLOWED_SEARCH_ENGINES	all engines	Comma-separated allow-list
BRAVE_API_KEY	—	Required for Brave search
GITHUB_TOKEN	—	Optional, raises GitHub API rate limit
USE_PROXY	false	Enable HTTP/SOCKS5 proxy
PROXY_URL	http://127.0.0.1:7890	Proxy address

MCP Tools

web_search

Search the web using one or more engines simultaneously.

{
  "query": "rust async runtime",
  "engines": ["bing", "duckduckgo"],
  "limit": 10
}

Parameter	Type	Default	Description
query	string	required	Search query
engines	string[]	server default	Engines to search
limit	number	10	Max results per engine (1–50)

Supported engines: bing, duckduckgo, brave, baidu, csdn, juejin, zhihu, github, github_code

fetch_web_content

Fetch and extract text content from any public URL.

Parameter	Type	Default	Description
url	string	required	Public HTTP/HTTPS URL
max_chars	number	30000	Max characters to return

fetch_github_readme

Fetch the README from a GitHub repository.

Parameter	Type	Description
url	string	GitHub repository URL

fetch_csdn_article / fetch_juejin_article / fetch_zhihu_article

Domain-specific fetchers with better content extraction.

Tool	URL Constraint
fetch_csdn_article	must contain csdn.net
fetch_juejin_article	must contain juejin.cn and /post/
fetch_zhihu_article	must contain zhihu.com

Development

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run with MCP Inspector for debugging
mcp dev src/mcp_ferris_search/server.py

Publishing to PyPI

# Build
python -m build

# Upload
python -m twine upload dist/*

Deploying to ModelScope MCP Hub

1. Publish to PyPI first
2. Go to https://modelscope.cn/mcp/servers/create?template=customize
3. Fill in the form:
   • Name: Ferris Search
   • Source: PyPI package mcp-ferris-search
   • Hosting Type: 可托管部署 (Hosted)
   • Command: uvx mcp-ferris-search

License

Apache-2.0

Acknowledgements

• Inspired by ferris-search (Rust version)
• Built with FastMCP and Model Context Protocol