Model Context Protocol
An open standard that connects AI applications to tools and data through servers that any compatible host can use.
In more detail
A host creates one client per server; servers offer tools for the model, resources for the application and prompts for the user, over JSON-RPC 2.0. The 2026-07-28 revision made the protocol stateless: no initialize handshake and no sessions, with each request carrying its protocol version and client capabilities and server/discover reporting what a server supports.
Related terms
- MCP host — The AI application a user works in, such as a chat app or IDE, which creates MCP clients and owns the conversation.
- MCP client — The component inside a host that connects to one MCP server, discovers it, and sends it requests on the host's behalf.
- MCP server — A program that exposes tools, resources or prompts to AI applications through the Model Context Protocol.
- JSON-RPC — A lightweight protocol for remote calls encoded as JSON objects, with requests, results, errors and notifications matched by id.
- server/discover — The MCP method, new in the 2026-07-28 revision, that returns the protocol versions, capabilities and identity a server supports.
Taught in
- Agent SDKs And Human Approval — AI Engineering: LLM Apps, RAG & Agents
- What MCP Is — AI Engineering: LLM Apps, RAG & Agents
- Build An MCP Server — AI Engineering: LLM Apps, RAG & Agents
- Connect An Agent To MCP — AI Engineering: LLM Apps, RAG & Agents
- Remote Servers And Trust — AI Engineering: LLM Apps, RAG & Agents
- Working With AI Coding Assistants — AI Engineering: LLM Apps, RAG & Agents
- Capstone: The Tasks Assistant — AI Engineering: LLM Apps, RAG & Agents