---
published: true
id: mcp
title: MCP
description: Knowledge captured in João Lopes' learning graph.
tags:
- ai-engineering
- mcp
- claude-code
- protocol
locale: en
resource: https://joaolopes.dev.br/knowledge/concepts/en/mcp/
created: '2026-03-19'
---

# MCP
Model Context Protocol, an open standard JSON-RPC interface that lets AI agents discover and call external tools through a unified protocol.

## Definition
MCP defines a standardized way for AI agents to interact with external tools. An MCP server exposes tools with names, descriptions, and parameter schemas, and any MCP-compatible client (Claude Code, Cursor, VS Code) can use them. Each tool is self-describing, enabling capability discovery at runtime and cross-client portability. Key tradeoffs include context cost (MCP responses are often verbose JSON consuming many tokens) and blocking behavior (each call blocks the agent until it returns, with no parallelism).

## Source
- [Model Context Protocol — Official Site](https://modelcontextprotocol.io/)
- [MCP Specification](https://modelcontextprotocol.io/specification/2025-11-25)
- [Model Context Protocol — Wikipedia](https://en.wikipedia.org/wiki/Model_Context_Protocol)

## Related
- [JSON](/concepts/en/json.md)
- [Model](/concepts/en/model.md)