MCP Agents Cloud
- other
MCP Agents Cloud是一个提供管理MCP服务器和代理（Agent）的平台。
About
What is MCP Agents Cloud?
MCP Agents Cloud is a self-hosted platform for managing MCP servers and agents. It provides user management, agent SDK (Python package and HTTP API), and streaming chat interactions. Built with Python FastAPI and Vue.js, it requires MySQL 8.0 and is intended for developers or teams who need to orchestrate MCP-based agents.
How to use MCP Agents Cloud?
Install MySQL via Docker (docker-compose up mysql in the docker directory). Set up the backend by creating a Python virtual environment, installing dependencies, copying a YAML configuration file, and running uvicorn main:app --reload. Set up the frontend with npm install then npm run serve for development or npm run build for production. Configuration is managed through environment‑specific YAML files (config-dev.yml, config-test.yml, config-prod.yml).
Key features of MCP Agents Cloud
- User management (registration, login, password reset)
- MCP server CRUD operations
- Agent CRUD and management
- Agent SDK with Python package and HTTP API
- Streaming chat interaction with agents
- SDK API key management
Use cases of MCP Agents Cloud
- Centralized management of multiple MCP servers and configurations
- Creating, editing, and deploying AI agents powered by OpenAI models
- Integrating agents into external applications via the provided SDK
- Team‑based collaborative development of MCP agents with user roles
FAQ from MCP Agents Cloud
What is the difference between MCP Agents Cloud and directly using the MCP protocol?
MCP Agents Cloud adds a management layer with a web UI, user authentication, and an agent SDK. It handles server and agent lifecycle, making it easier to coordinate multiple MCP resources without building those features from scratch.
Which AI models does MCP Agents Cloud support?
The platform’s backend uses the openai-agent framework and includes OpenAI API configuration (key and model settings). It therefore supports OpenAI models as the primary inference engine.
Is MCP Agents Cloud free and open source?
Yes, the project is released under the MIT license. You can deploy it yourself on any infrastructure that meets the requirements (MySQL, Python, Node.js).
How do I set up the required database?
MCP Agents Cloud requires MySQL 8.0. A Docker Compose file is provided in the docker directory – run docker-compose up mysql to start a MySQL instance on port 63307. The data directory (/appdata/share/agents-mcp-cloud/mysql/data) must exist and be owned by your user.
Does MCP Agents Cloud come with API documentation?
Yes. After starting the backend server, API documentation is available at http://localhost:8000/docs (Swagger UI). The configuration also has an api.docs_url setting.
Details
- Author
- hicodermonkey
- GitHub stars
- 3
- Category
- other
- Repository
- hicodermonkey/agents-mcp-cloud
MCP Agents Cloud
MCP Agents Cloud是一个提供管理MCP服务器和代理（Agent）的平台。
功能特性
- 用户管理（注册、登录、密码重置）
- MCP服务器管理（添加、编辑、删除、查看）
- Agent管理（创建、编辑、删除、查看）
- Agent SDK（Python包和HTTP API）
- Agent聊天交互（流式输出）
- SDK密钥管理
技术栈
前端
- Vue.js - Element UI - Axios - Vue Router - Vuex后端
- Python - FastAPI - SQLAlchemy - JWT - openai-agent框架数据库
- MySQL 8.0快速开始
安装依赖组件
项目依赖MySQL数据库，可以通过Docker快速安装：
# 进入docker目录
cd docker
确保数据目录存在
sudo mkdir -p /appdata/share/agents-mcp-cloud/mysql/data
sudo chown -R $USER:$USER /appdata/share/agents-mcp-cloud/mysql/data
启动MySQL服务
docker-compose up -d mysql
数据库将在63307端口启动。详细说明请参考Docker依赖组件文档。
后端设置
1. 进入后端目录：
cd backend
2. 创建虚拟环境：
python -m venv venv
3. 激活虚拟环境：
# Windows
venv\Scripts\activate
Linux/Mac
source venv/bin/activate
4. 安装依赖：
pip install -r requirements.txt
5. 配置系统：
# 根据环境选择复制对应的配置文件
cp config.yml.example config-dev.yml
然后编辑配置文件
6. 运行应用：
# 设置环境（默认为dev）
export ENV=dev # 或 test, prod
或直接指定配置文件的绝对路径
export CONFIG_FILE="$(pwd)/config-dev.yml"
运行服务
cd backend # 确保在backend目录下运行
uvicorn main:app --reload
前端设置
1. 进入前端目录：
cd frontend
2. 安装依赖：
npm install
3. 运行开发服务器：
npm run serve
4. 构建生产版本：
npm run build
配置说明
项目使用基于环境的YAML配置文件：
- config-dev.yml：开发环境配置
- config-test.yml：测试环境配置
- config-prod.yml：生产环境配置
配置文件应放在backend目录下，可以通过以下两种方式指定配置：
# 1. 通过环境变量指定环境，系统将查找backend/config-{ENV}.yml
export ENV=dev # 或 test, prod
2. 直接指定配置文件的绝对路径（推荐）
export CONFIG_FILE="/完整路径/到配置文件/config-dev.yml"
主要配置项包括：
- app: 应用名称、版本等基本信息
- api: API路径前缀和文档URL
- server: 服务器主机、端口设置
- security: 安全设置（密钥、令牌过期时间）
- cors: 跨域设置
- database: 数据库连接URL和连接池设置
- openai: OpenAI API密钥和模型设置
- email: 电子邮件发送设置
- smtp: SMTP服务器配置
详细配置示例请参考config.yml.example文件。
API文档
启动后端服务器后，可以在以下地址访问API文档：
http://localhost:8000/docs
项目结构
agents-mcp-cloud/
├── backend/ # 后端代码
│ ├── app/ # 应用代码
│ │ ├── api/ # API路由
│ │ ├── core/ # 核心功能
│ │ ├── db/ # 数据库工具
│ │ ├── models/ # 数据模型
│ │ ├── schemas/ # Pydantic模式
│ │ ├── utils/ # 工具函数
│ │ └── services/ # 业务服务
│ ├── config-dev.yml # 开发环境配置
│ ├── config-test.yml # 测试环境配置
│ ├── config-prod.yml # 生产环境配置
│ ├── main.py # 主应用入口
│ └── requirements.txt # 依赖列表
├── docker/ # Docker依赖组件
│ ├── docker-compose.yml # Docker编排文件
│ └── mysql/ # MySQL相关配置
├── frontend/ # 前端代码
│ ├── public/ # 静态资源
│ ├── src/ # 源代码
│ │ ├── assets/ # 资源文件
│ │ ├── components/ # Vue组件
│ │ ├── router/ # 路由配置
│ │ ├── store/ # Vuex存储
│ │ ├── views/ # 页面视图
│ │ └── api/ # API客户端
│ └── package.json # 依赖列表
└── scripts/ # 脚本文件
