This skill provides a comprehensive framework for building, maintaining, and scaling Model Context Protocol (MCP) servers. It offers domain-specific guidance on implementing tools, resources, and prompts, while ensuring data integrity through Zod validation. Whether you are setting up local stdio transports for Claude Desktop or implementing Streamable HTTP for remote clients like Cursor, this skill helps developers navigate the evolving SDK landscape and adhere to industry best practices for AI agent connectivity and tool execution.
主な機能
01Guidance on choosing between stdio and Streamable HTTP transports
02Best practices for idempotent tool design and structured error handling
03Robust input validation implementation using Zod schemas
04Reference templates for TypeScript and Node.js MCP SDK integration
05Standardized patterns for tool, resource, and prompt registration
06151 GitHub stars
ユースケース
01Building custom MCP servers to connect proprietary APIs to Claude Code
02Migrating local developer tools to remote cloud environments via Streamable HTTP
03Standardizing AI agent data retrieval through structured MCP resource handlers