What is Model Context Protocol (MCP)?
Model Context Protocol (MCP) is an open standard introduced by Anthropic in 2024 that lets AI assistants discover and use external tools — like USB-C for AI, providing a universal way to plug models into apps and data sources.
Before MCP, every AI assistant had to implement its own bespoke way of talking to external tools. ChatGPT had plugins, Claude had tool use, custom agents had custom integrations — each incompatible with the others. Developers had to rebuild the same integration five times for five different AI products.
MCP standardizes this. An MCP server exposes a set of tools (functions the AI can call), resources (data the AI can read), and prompts (templates the AI can use). Any MCP-compatible client — Claude Desktop, Cursor, Windsurf, ChatGPT, custom agents — can connect to any MCP server and immediately get access to its capabilities. Build once, use everywhere.
The protocol covers authentication, tool discovery, resource subscriptions, and streaming responses. It's designed for both local servers (running on the user's machine) and remote servers (hosted services). Adoption has been fast: by 2026, most major AI platforms support MCP as a first-class way to extend their capabilities.
How Definable uses MCP
Definable runs an MCP server that exposes Definable's 1000+ app integrations, knowledge bases, and workflow capabilities to any MCP-compatible client. Use it from Claude Desktop, Cursor, or any custom agent. Definable Assistant also acts as an MCP client, consuming third-party MCP servers to extend what it can do.
Frequently asked questions
Who created MCP?
Anthropic open-sourced the Model Context Protocol in late 2024. It's now widely adopted across the industry as the standard way for AI clients to talk to external tools.
Is MCP only for Claude?
No. MCP is provider-agnostic. Any AI system can speak the protocol — Claude, GPT, Gemini, and custom models all support it via clients or libraries.
Do I need an MCP server to use AI tools?
Not necessarily. Many AI platforms include built-in tool integrations. MCP becomes useful when you want to expose your own custom tools, or when you want one set of tools to work across many AI clients.
Try Definable AI free
50+ AI models, 1000+ integrations, knowledge bases, photo & video studios — all in one platform.