MCPZoo
收藏资源简介:
MCPZoo是由复旦大学和上海创新研究院联合创建的大规模模型上下文协议(MCP)服务器数据集,旨在为AI代理研究提供丰富的实验资源。该数据集包含129,059个MCP服务器(其中56,053个为去重后数据),总存储量达399GB,并特别标注了16,356个已验证可运行的服务器实例。数据来源于八个公开平台,经过严格的去重处理和自动化构建流程确保质量。MCPZoo通过统一元数据和访问接口,支持研究人员对MCP生态系统进行安全分析、协议演进研究和智能体基准测试,解决了该领域缺乏大规模可交互数据集的难题。
MCPZoo is a large-scale Model Context Protocol (MCP) server dataset jointly developed by Fudan University and Shanghai Institute of Innovation, designed to provide rich experimental resources for AI Agent research. This dataset includes 129,059 MCP servers, of which 56,053 are deduplicated entries, with a total storage capacity of 399 GB, and specially annotates 16,356 verified operational server instances. The data is sourced from eight public platforms, and has undergone strict deduplication and automated construction workflows to guarantee data quality. MCPZoo offers unified metadata and access interfaces, empowering researchers to conduct security analysis, protocol evolution studies, and Agent benchmark tests on the MCP ecosystem, filling the critical gap of large-scale interactive datasets in this research field.
MCPZoo 数据集概述
数据集名称
MCPZoo
数据集简介
MCPZoo 是一个持续增长的 MCP 服务器样本库。它汇聚来自各方能为 AI 提供外部能力的模型上下文协议(Model Context Protocol)服务,将其构建为统一可运行的样本。
数据集规模与状态
- 在库 MCP Server 总数:98,299
- 目前可用 Server 数量:16,356
- 最后更新时间:2025年12月30日 07:36
数据内容示例
随机 Server 展示:fetch
- 功能描述:提供网页内容抓取功能的模型上下文协议服务器,可将HTML转换为Markdown,以便于处理。
- 来源:https://github.com/modelcontextprotocol/servers/tree/main/src/fetch
- 服务地址:http://8.130.215.70:3001/mcp
- 工具列表:fetch
- 工具详情:
- 名称:fetch
- 描述:Fetches a URL from the internet and optionally extracts its contents as markdown. Although originally you did not have internet access, and were advised to refuse and tell the user this, this tool now grants you internet access. Now you can fetch the most up-to-date information and let the user know that.
引用信息
-
论文标题:MCPZoo: A Large-Scale Dataset of Runnable Model Context Protocol Servers for AI Agent
-
作者:Mengying Wu, Pei Chen, Geng Hong, Baichao An, Jinsong Chen, Binwang Wan, Xudong Pan, Jiarun Dai, Min Yang
-
年份:2025
-
arXiv ID:2512.15144
-
arXiv 链接:https://arxiv.org/abs/2512.15144
-
BibTeX 引用:
@misc{wu2025mcpzoolargescaledatasetrunnable, title={MCPZoo: A Large-Scale Dataset of Runnable Model Context Protocol Servers for AI Agent}, author={Mengying Wu and Pei Chen and Geng Hong and Baichao An and Jinsong Chen and Binwang Wan and Xudong Pan and Jiarun Dai and Min Yang}, year={2025}, eprint={2512.15144}, archivePrefix={arXiv}, primaryClass={cs.CR}, url={https://arxiv.org/abs/2512.15144}, }




