MCPCorpus는 Model Context Protocol (MCP) 생태계를 분석하기위한 포괄적 인 데이터 세트이며, ~14K MCP 서버 및 20 + 정상화 된 메타 데이터 속성이있는 300 MCP 클라이언트가 포함되어 있습니다.

Dataset 개요

• Scale: ~14K MCP 서버 + 300 MCP 클라이언트
• Attributes: artifact 당 20+ 정상화된 메타데이터
• 소스: GitHub, 커뮤니티 허브, 패키지 관리자 -Applications: 생태계 분석, 보안 연구, 상호 운용성 연구

구조

⭐ 빠른 시작

데이터셋을 탐험

접근 데이터 프로그래밍

갱신 Dataset (선택)

당신이 당신의 연구에 있는 MCPCorpus를 사용하는 경우에, 그것을 인용하십시오:

MCPCorpus: A Large-Scale Evolvable Dataset for Model Context Protocol Ecosystem and Security Analysis

MCPCorpus is a comprehensive dataset for analyzing the Model Context Protocol (MCP) ecosystem, containing ~14K MCP servers and 300 MCP clients with 20+ normalized metadata attributes.

📊 Dataset Overview

• Scale: ~14K MCP servers + 300 MCP clients
• Attributes: 20+ normalized metadata per artifact
• Sources: GitHub, community hubs, package managers
• Applications: Ecosystem analysis, security research, interoperability studies

📁 Structure

MCPCorpus/
├── Crawler/                    # Data collection tools
│   ├── Servers/               # Server data
│   ├── Clients/               # Client data
│   ├── github_info_collector.py  # GitHub metadata collector
|   ├── data_cleaner.py        # Data normalization 
│   └── tool_extractor.py      # mcp tool extract
└── Website/                   # Web search interface
    ├── server.py              # Local web server
    └── index.html             # Search interface
    └── mcpso_servers_cleaned.json
    └── mcpso_clients_cleaned.json

🚀 Quick Start

Explore Dataset

cd Website
python server.py
# Open http://localhost:8000

Access Data Programmatically

import json
import pandas as pd

# Load datasets
with open('Crawler/Servers/mcpso_servers_cleaned.json', 'r') as f:
    servers = json.load(f)
with open('Crawler/Clients/mcpso_clients_cleaned.json', 'r') as f:
    clients = json.load(f)

# Convert to DataFrame
servers_df = pd.DataFrame(servers)
clients_df = pd.DataFrame(clients)

Update Dataset (Optional)

# Collect new data
cd Crawler/Servers && python Server_request.py
cd ../Clients && python Client_request.py

# Add GitHub metadata
cd .. && python github_info_collector.py --token YOUR_GITHUB_TOKEN

📚 Citation

If you use MCPCorpus in your research, please cite it as:

@misc{lin2025largescaleevolvabledatasetmodel,
      title={A Large-Scale Evolvable Dataset for Model Context Protocol Ecosystem and Security Analysis}, 
      author={Zhiwei Lin and Bonan Ruan and Jiahao Liu and Weibo Zhao},
      year={2025},
      eprint={2506.23474},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      url={https://arxiv.org/abs/2506.23474}, 
}