Concepts
MCP, explained
Plain-language answers to the questions everyone evaluating MCP asks — what the protocol is, what a server actually contains, how it differs from the API you already run, and where skills fit.
What is MCP (Model Context Protocol)?MCP (Model Context Protocol) is the open standard that lets AI models discover and use external tools at runtime — and every major AI client speaks it.What is an MCP server?An MCP server exposes typed tools that Claude, ChatGPT, and Cursor call at runtime — what one contains, local vs. remote, and what production takes.MCP vs. API: what’s the difference?APIs serve developers who integrate ahead of time; MCP serves agents that discover tools at runtime — what changes when an API becomes an MCP server.What is an AI skill?An AI skill is a versioned package of instructions and guardrails that teaches an agent to do a job well — competence, where MCP tools give capability.What is Code Mode?Code Mode lets an agent write JavaScript that calls your MCP tools in a sandbox, so intermediate payloads never enter its context window.How do my customers sign in to my MCP server?Four ways a publisher can identify the people calling their MCP server: an emailed code, your OAuth provider, your own sign-in, or an MCPCloud account.