# Model Context Protocol (MCP)

> A shared plug that lets any AI app use tools and data.

**Field:** Artificial Intelligence  
**Category:** MCP  
**Also known as:** MCP, Model Context Protocol

## Think of it like

Think of it like one plug that fits every socket. Any app can use any tool that has it.

## Definition

The Model Context Protocol, or MCP, is an open standard for linking AI apps to tools and data. A tool maker builds one MCP server, and any app that speaks MCP can use it. Before MCP, each app needed its own custom link to each tool. The protocol sets how the two sides find out what the other offers and how they swap messages. It was first released in 2024 and is now used by most major AI apps.

## Example

Lucy adds one GitHub server to her settings. Both her chat app and her code editor can now read her issues, with no extra code.

## Example code

```json
{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"]
    }
  }
}
```

A typical settings entry that adds one MCP server. The app starts it and finds its tools.

## Why it matters

It turns many one-off links into one shared plug. A tool built once works in every app.

## Common mistake

Thinking MCP is a model or an app. It is only the agreement on how apps and tools talk.

## Three principles

- **One shared plug.** Build once, use anywhere.
- **Apps and tools.** Two sides, one language.
- **Open standard.** Not tied to one maker.

## Related terms

- [MCP Server](https://learn.benibix.com/term/ai/mcp-server)
- [MCP Client](https://learn.benibix.com/term/ai/mcp-client)
- [Tool Use (Function Calling)](https://learn.benibix.com/term/ai/tool-use)
- [JSON-RPC](https://learn.benibix.com/term/development/json-rpc)
- [MCP Tool Search](https://learn.benibix.com/term/ai/mcp-tool-search)
- [Tool Poisoning](https://learn.benibix.com/term/ai/tool-poisoning)
- [MCP Protocol Version](https://learn.benibix.com/term/ai/mcp-protocol-version)

## Told apart from

- [MCP vs Tool Use](https://learn.benibix.com/compare/model-context-protocol-vs-tool-use): People often think MCP is a new way for models to use tools that replaces tool use. It is not. Tool use is what the model does: it asks for a tool by name and reads the result. MCP is how an app finds tools made by others and plugs them in. Under MCP, the model still uses tools in the same way.
- [A2A vs MCP](https://learn.benibix.com/compare/agent2agent-protocol-vs-model-context-protocol): People often think A2A and MCP are rival standards that do the same job. MCP connects one agent to its tools and data, such as files or a database. A2A connects one agent to other agents, so they can hand each other tasks. Many systems use both: MCP inside each agent, and A2A between them.

![Three AI apps on one side and three tools on the other: files, a database and a calendar. Without MCP, each app needs its own link to each tool. With MCP, all of them plug into one shared protocol.](https://learn.benibix.com/diagrams/mcp-protocol.svg)

---

Source: https://learn.benibix.com/term/ai/model-context-protocol
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
