What problem does it solve? Building a Model Context Protocol (MCP) server that works across Claude Desktop, ChatGPT, Cursor, and other AI hosts requires navigating platform-specific transport rules, JSON-RPC handshake requirements, and context window limits, which this Skill codifies into a repeatable engineering workflow. ## Core Features & Use Cases - Automated Scaffolding: Generate complete MCP server projects via scripts/mcp_scaffold.py with archetypes for universal dual-transport servers, multi-server gateways, TypeScript stdio/HTTP, Python FastMCP, private-API portals, and Puppeteer browser automation. - Protocol Verification: Run scripts/verify_mcp_server.py to test the initialize handshake, ping, tool discovery, and detect stdout pollution that corrupts stdio JSON-RPC streams. - Platform Compatibility Guidance: Reference playbooks cover ChatGPT's HTTPS-only and tools-only constraints, readOnlyHint requirements, search/fetch connector tools, and the 4-tool Meta-Tool gateway pattern that cuts context usage by roughly 97%. - Use Case: Connect an internal enterprise portal with no public API to Claude Desktop by scaffolding a private-api server, implementing session re-authentication, verifying the protocol handshake, and registering the host configuration. ## Quick Start Ask the agent to scaffold a universal MCP server for your target service and verify it with the protocol tester.