mcp-fast-deploy

> Note: This project is a Serverless Devs application. Since there are variables that need to be initialized before the application can run (such as deployment region, function name, etc.), it is **not recommended** to directly clone this repository to your local machine for deployment or to directly copy the s.yaml file. It is **strongly recommended** to initialize using the method `s init ${template name}` or through the application center. For more details, please refer to [Deployment & Experience](#deployment--experience). # start-mcp-fast-deploy Documentation <description> The STDIO MCP service transforms into an SSE remote service through one-click deployment with Function Compute FC. </description> ## Resource Preparation To use this project, you need to have the following services activated and possess the corresponding permissions: <service> | Service/Business | Permission | Related Documentation | | --- | --- | --- | | Function Compute | AliyunFCFullAccess | [Help Documentation](https://help.aliyun.com/product/2508973.html) [Billing Documentation](https://help.aliyun.com/document_detail/2512928.html) | | Log Service | AliyunFCServerlessDevsRolePolicy | [Help Documentation](https://help.aliyun.com/zh/sls) [Billing Documentation](https://help.aliyun.com/zh/sls/product-overview/billing) | </service> <remark> </remark> <disclaimers> </disclaimers> ## Deployment & Experience <appcenter> - :fire: Deploy this application through the [Cloud Native Application Development Platform CAP](https://cap.console.aliyun.com/template-detail?template=start-mcp-fast-deploy), [](https://cap.console.aliyun.com/template-detail?template=start-mcp-fast-deploy). </appcenter> <deploy> </deploy> ## Case Introduction <appdetail id="flushContent"> Currently, various MCP Servers available on the market are mostly based on STDIO implementations and packaged as standalone binary executable files for distribution. Among them, Node.js and Python dominate the landscape. The Node.js ecosystem typically provides packages via npx, for example, "npx -y @amap/amap-maps-mcp-server", while the Python ecosystem uses uvx, for example, "uvx mcp-server-time --local-timezone=Asia/Shanghai". Therefore, transforming mainstream community STDIO MCP Servers into your own MCP Server with one click is a very valuable endeavor. This case demonstrates how the STDIO MCP service can be transformed into an SSE remote service through one-click deployment with Function Compute FC. </appdetail> ## Architecture Diagram <framework id="flushContent">  </framework> ## Usage Process <usedetail id="flushContent"> After successful deployment: 1. Open this web application, and you can choose to deploy using npx or uvx. a. We will take Gaode Map @amap/amap-maps-mcp-server as an example. b. After filling in the details, click the "Submit Deployment" button. After a moment, it will display a successful deployment message along with the MCP service URL and the function ARN that supports this MCP service.  2. Directly use the "npx @modelcontextprotocol/inspector" debugger to debug the successfully deployed MCP SSE service.  3. After that, you can register this URL with various Agent clients for consumption, such as BaiLian, Cherry Studio, etc. Taking Cherry Studio as an example:   </usedetail> ## Secondary Development Guide <development id="flushContent"> You can carry out secondary development based on https://github.com/devsapp/mcp-fast-deploy according to your needs. </development>