AI Agents and MCPStart here
- What is an AI agent? The loop of asking, running a tool and reading the result.
- What is MCP? A standard way to connect AI applications to tools and data.
- How MCP is built: hosts, clients and servers, and the three kinds of things a server offers.
- A tiny MCP server: the code and the real messages we recorded.
- Security basics: read before you connect anything to real data.
The message checker
MCP messages are JSON-RPC 2.0 messages. Paste one above to see what kind it is (request, notification, response or error) and whether its shape is right. It checks the shape only: it does not run anything and sends nothing anywhere.
Did you use an AI model for the demos?
No. The loop uses a labelled stand-in function. The server and the messages are real.
Is this a tutorial for a particular product?
No. It follows the open MCP documentation and Anthropic's tool use documentation.
- What is an AI agent?
- What is MCP (Model Context Protocol)?
- How MCP is built: hosts, clients, servers
- Build a tiny MCP server and see the real messages
- MCP security basics
This is an independent guide. It is not connected with Anthropic or the MCP project. This field changes quickly: statements were read on 14 November 2026 from the pages named under Sources (the MCP site shows specification version 2026-07-28 in its page addresses). The demo server and the messages are real: we ran them with the official Python SDK (mcp 2.3.0). No AI model was used for the demos.