Creativity · MCP — overview
What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard introduced by Anthropic in late 2024 that defines how LLM clients (Claude Desktop, IDEs, agent frameworks) connect to external tools, resources, and prompts through uniform MCP servers. Instead of bespoke integrations per model and per tool, any compliant client can speak to any compliant server.
MCP facts
- Kind
- overview
- Ecosystem
- anthropic-mcp
- Transports
- stdio, sse, http
Capabilities
- Tools — typed functions the model can invoke
- Resources — data the client can expose read-only (files, DB rows, URLs)
- Prompts — reusable prompt templates the server can offer the client
- Sampling — the server can request LLM completions from the client
Frequently asked questions
What is MCP?
The Model Context Protocol (MCP) is an open standard for connecting LLM applications to external tools, data, and prompts. Introduced by Anthropic in November 2024, it defines a uniform client-server interface so any compliant client can use any compliant server.
What problem does MCP solve?
MCP replaces N×M bespoke integrations between LLM clients and tools with one protocol. A single MCP server for, say, your filesystem or database, works with Claude Desktop, Cursor, Zed, and any other compliant client — without per-client glue.
Is MCP only for Claude?
No. MCP is an open specification. Anthropic introduced it, but clients and servers exist across the ecosystem — Cursor, Zed, Windsurf, LangChain, LlamaIndex adapters, and community servers numbering in the hundreds.
What are the three MCP primitives?
Tools (typed functions the model can call), Resources (read-only data the model can access), and Prompts (reusable templates the server offers to the client). Sampling is a fourth capability — the server can request completions from the client's model.
Sources
- Model Context Protocol — official spec — accessed 2026-04-20
- Anthropic — MCP announcement — accessed 2026-04-20