What is MCP (Model Context Protocol)?

MCP (Model Context Protocol) is an open standard that lets AI assistants like Claude, ChatGPT, Cursor and Codex use outside tools and data. An MCP server is a small program that exposes tools, for example "search GitHub issues" or "query this database", which the assistant can call when it needs them.

Why it exists

Before MCP, every AI app needed its own custom integration for every tool. MCP gives them one shared way to connect, so a server written once works in many assistants.

What an MCP server provides

Tools: actions the assistant can run, like creating a ticket or reading a file.
Resources: data the assistant can read, like documents or database rows.
Prompts: ready-made instructions for common tasks.

Local and remote servers

A local server runs on your computer, usually started with npx or uvx. It can reach your files and local apps. A remote server is hosted on the internet and you connect with a URL. Remote servers work in more places, including ChatGPT, which cannot run local servers.

Is it safe?

An MCP server can do whatever its tools allow, so treat it like any software you install: check who made it, whether it is maintained, and what access it asks for. See are MCP servers safe.

Questions

Who created MCP?

Anthropic introduced MCP as an open standard in November 2024. It is now supported by Claude, ChatGPT, Cursor, VS Code, Codex, Gemini CLI and many other clients.

Do I need to code to use an MCP server?

No. Most servers install with one command or one block of settings. Every MCP server page on this site shows the exact command for each assistant.