awslabs/mcp — reverse-engineered prompt

Build me an open source Python project that gives AI coding assistants access to AWS through the Model Context Protocol. I want it to be a collection of small MCP servers that people can turn on depending on what they are building, like AWS docs lookup, infrastructure help, serverless guidance, containers, databases, analytics, AI services, costs, support, and developer workflows.

Make it work with stdio MCP clients and include clear setup instructions for tools like Cursor, Claude Code, VS Code, Windsurf, Kiro, Cline, and Claude Desktop. The goal is that a developer can install one of the servers, add a simple config file to their AI tool, and then ask better AWS questions or get more useful help while coding.

Please include practical samples, good docs, security notes about credentials and secrets, and enough tests so contributors can add more AWS MCP servers later. Look up current MCP and AWS docs online if you need to.