Start here
MCP is the protocol an AI client uses to call tools that run somewhere else. These pages define the words the rest of the tutorials assume: server, client, tool, agent, and how all of it differs from an ordinary API. Read them first if any of those are new.
Start here: the basics
- What is an MCP server?An MCP server gives AI apps like Claude and ChatGPT a list of tools they can use in another system, such as Jira or GitHub. What MCP is, what a server and a tool are, and what happens when an agent calls one.
- MCP vs APIs: what is the difference?An API is how software talks to a service. An MCP server usually sits on top of an API and describes it as tools an AI agent can find and call. Diagram, table and examples.
- What is an AI agent?An AI agent is an AI model that works toward a goal by choosing steps and using tools, like running a search or editing a file. Real examples, and what each one is allowed to touch.
- AI agents vs agentic AI: what is the difference?An AI agent is one system that works toward a goal using tools. Agentic AI is the broader label for AI built to act on its own. The difference, with an example.
- What is an LLM gateway?An LLM gateway is a proxy that sits between your app or agent and the model providers behind it, so you call one API instead of OpenAI's, Anthropic's and Google's separately. What it routes, caches, logs and blocks, with quotes from LiteLLM, Portkey, Cloudflare and Kong.
- What is an agent harness?An agent harness is the software around a model that turns it into an agent: the loop, tool calling, context management, memory, permissions and the system prompt. How the term differs from an evaluation harness.
- What is an MCP client?An MCP client is the part of an AI app that holds one connection to one MCP server, runs the handshake, and can offer sampling, roots or elicitation.
- MCP vs function calling: what's the difference?Function calling is how a model API describes and returns tool calls. MCP is a protocol that supplies the tool definitions and runs the calls.
- MCP vs A2A: what is the difference?MCP connects an AI agent to tools and data. A2A connects one agent to another, independent agent. What each protocol covers, and when a system needs both.
- MCP stdio vs Streamable HTTP: what is the difference?MCP's two transports, stdio and Streamable HTTP, on the wire: the real messages each sends, what the 2026-07-28 spec changed, and their security rules.
- MCP resources and prompts: who controls each oneTools aren't all an MCP server offers. The spec adds resources and prompts, and puts each of the three under a different owner: model, app or user.