One‑Click Migration of STDIO MCP Servers to Serverless SSE Services

Why Function Compute for MCP Servers

Function Compute provides secure, cost‑effective, and elastic runtime for MCP services, making it suitable for enterprise deployments.

One‑Click Conversion of STDIO MCP Servers

Use the FunctionAI template

https://cap.console.aliyun.com/create-project?template=start-mcp-fast-deploy

to create a web app that packages the chosen MCP server (e.g., @amap/amap-maps-mcp-server) and provisions a Function Compute instance.

After deployment, obtain the web app URL, which includes the MCP service endpoint and the function ARN.

Debug the deployed server with npx @modelcontextprotocol/inspector.

Register the SSE URL in MCP clients such as Cherry Studio.

The deployment uses the mcp-helper function to download dependencies, zip them, upload to OSS, and create the Function Compute instance with the appropriate start command.

Transforming Existing OpenAPI Services into MCP Servers

Use the FunctionAI template

https://cap.console.aliyun.com/create-project?template=start-fcai-mcp-openapi

and provide an OpenAPI JSON spec plus optional security headers.

Supported authentication methods are API Key, HTTP Basic, and OAuth 2.0; each method has example configuration snippets.

After deployment, the generated SSE URL can be consumed by MCP clients (e.g., Cursor, Cherry Studio).

Optionally supply a custom MCP configuration JSON that maps service names to URLs.

OpenAPI‑to‑MCP Conversion Library

The open‑source library @serverless-devs/openapi-mcp-converter converts OpenAPI definitions into MCP server instances. Example usage:

import fs from 'fs';
import { OpenApiMCPSeverConverter } from '../index.js';
import { StdioServerTransport } from '@modelcontextprotocol/sdk/server/stdio.js';
import path from 'path';
import { fileURLToPath } from 'url';

const openApiDoc = JSON.parse(
  fs.readFileSync(
    path.join(path.dirname(fileURLToPath(import.meta.url)), 'openapi.json'),
    'utf8'
  )
);

const converter = new OpenApiMCPSeverConverter(openApiDoc, {
  timeout: 100000,
  security: { apiKey: 'my-api-key' }
});
const server = converter.getServer();

Repository: https://github.com/Serverless-Devs/openapi-mcp-converter

FAQ

"s.yaml not found in response" – ensure the OpenAPI spec does not contain double curly braces {{}}.

"Status Code is not 200" – verify the servers field is non‑empty and points to a reachable endpoint.

Conclusion

Combining a protocol‑translation adapter with Function Compute enables enterprise‑grade, low‑cost deployment of MCP servers. It supports zero‑code migration of existing STDIO or OpenAPI services to SSE‑based MCP endpoints, facilitating rapid integration into AI ecosystems and MCP marketplaces.