A Large-Scale Dataset of MCP Implementations on GitHub
收藏资源简介:
该数据集由奥克兰大学和北卡罗来纳大学威明顿分校的研究团队构建,是一个大规模、基于证据的GitHub上模型上下文协议(MCP)实现的数据集。它包含2,297个经过验证的MCP项目,通过混合管道从GitHub REST和GraphQL API收集,并经过多阶段验证,涵盖客户端、服务器和网关三种操作角色,数据以JSONL模式组织。数据集创建过程涉及初始仓库发现、活跃维护过滤、多阶段验证管道以及手动验证与规则细化,确保高精度和可重复性。该数据集旨在支持对现实世界MCP生态系统的研究,应用于软件工程、数据挖掘和机器学习领域,帮助解决协议实现、集成、兼容性和演化分析等问题。
This dataset was constructed by a research team from the University of Auckland and the University of North Carolina Wilmington. It is a large-scale, evidence-based dataset of Model Context Protocol (MCP) implementations hosted on GitHub. The dataset contains 2,297 validated MCP projects, which were collected via a hybrid pipeline from GitHub REST and GraphQL APIs and subjected to multi-stage validation. It covers three operational roles: client, server, and gateway, and is organized in JSONL format. The dataset creation process involves initial repository discovery, active maintenance filtering, multi-stage validation pipelines, manual validation and rule refinement, ensuring high accuracy and reproducibility. This dataset is designed to support research on real-world MCP ecosystems, with applications in software engineering, data mining, and machine learning, to aid in addressing issues including protocol implementation, integration, compatibility, and evolutionary analysis.





