MCP Transport
MCP transports define how AI clients communicate with MCP servers. The protocol supports two transport types: HTTP (for cloud-deployed servers accessible over the internet) and STDIO (for local servers that communicate through standard input/output).
Explanation
Transport choice affects deployment, latency, and accessibility. HTTP transport makes your MCP server available to any AI client over the internet — ideal for SaaS products and public-facing services. It supports TLS encryption, standard authentication headers, and load balancing. STDIO transport runs your MCP server as a local process, communicating through stdin/stdout — ideal for developer tools, CLI integrations, and local-first workflows. Some MCP servers support both transports simultaneously, allowing the same codebase to serve both cloud and local clients. xmcp handles transport configuration automatically — you deploy to Vercel for HTTP or run locally for STDIO without changing your tool definitions.
Related Terms
Model Context Protocol (MCP)
Model Context Protocol (MCP) is an open standard that defines how AI assistants connect to external tools, data sources, and services. It provides a structured way for AI models to discover, understand, and use capabilities from any MCP-compatible server.
MCP Server
An MCP server is a service that implements the Model Context Protocol to expose tools, resources, and prompts to AI assistants. It handles connection management, capability negotiation, and request routing between AI clients and your underlying systems.
MCP Client
An MCP client is an AI assistant or application that connects to MCP servers to discover and use their capabilities. Popular MCP clients include Claude, Cursor, GitHub Copilot, and other AI-powered developer tools.
HTTP Transport
HTTP transport allows MCP servers to communicate with AI clients over standard HTTP connections. This is the recommended transport for production MCP servers that need to be accessible from cloud-hosted AI assistants and services.
STDIO Transport
STDIO transport allows MCP servers to run as local processes that communicate with AI clients through standard input (stdin) and standard output (stdout). This is ideal for developer tools, CLI integrations, and local-first AI workflows.
Ready to build your own MCP server?
Take the 2-minute quiz to find out if MCP is the right fit for your project, then get a step-by-step plan to ship it.