¿Qué es Model Context Protocol (MCP)? Definición y Contexto
MCP is an open protocol from Anthropic that lets language models connect to external tools and data sources through a standard interface — the USB-C of LLM tool use.
Model Context Protocol (MCP) is an open standard, introduced by Anthropic in late 2024, that lets language models connect to external tools, file systems, APIs, and data sources through a uniform JSON-RPC interface. Where every assistant used to define its own tool format, MCP gives them a shared protocol — write the integration once, plug it into any compliant agent.
A typical MCP server exposes three things: tools (callable functions), resources (readable content), and prompts (reusable templates). Clients — Claude Desktop, Cursor, Cline, custom agents — discover what a server offers and call into it without bespoke code.
The practical impact is that "give my agent access to X" goes from days of glue code to minutes of config. Most popular dev tools (filesystems, databases, GitHub, Slack, search engines) now ship MCP servers either officially or via the community.