Model Context Protocol (MCP)
An open-source standard for connecting AI applications to external systems — data sources, tools, and workflows. Developed by Anthropic and now broadly adopted across the AI ecosystem.
What It Is
MCP standardizes how AI applications connect to external context providers. The analogy used in the official docs: USB-C for AI apps — one universal interface instead of a custom connector per system.
Without MCP, every integration between an AI app and an external system (a database, a file system, a SaaS API) requires custom glue code on both sides. MCP replaces that with a single protocol both sides implement once.
Ecosystem Support
MCP is not Anthropic-only. Broad adoption:
| Category | Examples |
|---|---|
| AI assistants | Claude (Desktop, Code), ChatGPT |
| Dev tools / IDEs | VS Code (Copilot), Cursor, MCPJam |
| Remote servers | Sentry, filesystem, databases |
Build a server once; it works across all MCP-compatible clients.
What It Enables
- Agents accessing Google Calendar, Notion, databases — acting as personalized assistants
- Claude Code generating a web app from a Figma design
- Enterprise chatbots querying multiple databases via chat
- AI models controlling external tools (3D printers, Blender, etc.)
Scope
MCP defines only the protocol for context exchange — it does not dictate how AI applications use LLMs or manage the provided context. The ecosystem includes:
- MCP Specification — formal implementation requirements
- MCP SDKs — language-specific implementations
- MCP Inspector — development and debugging tool
- Reference server implementations
See Also
- mcp-architecture — hosts, clients, servers, transport, lifecycle
- mcp-primitives — tools, resources, prompts, and client-side primitives