The Model Context Protocol (MCP) is an open standard for connecting AI applications to external systems: data sources such as local files and databases, tools such as search engines and calculators, and workflows such as specialised prompts. An application that speaks the protocol reaches any compliant system through one interface instead of a bespoke integration per system.
This repository holds the protocol specification, the protocol schema, and the official documentation. The schema is defined in TypeScript first and published as JSON Schema as well for wider compatibility. The documentation is built with Mintlify and served at modelcontextprotocol.io.
The protocol covers both sides of a connection and the applications built on it: servers expose data and tools, client applications connect to those servers, and MCP apps run interactively inside an AI client.
Features
- Open standard: one connection model between AI applications and external data, tools, and workflows
- Specification and schema: the protocol is defined in TypeScript and also published as JSON Schema
- Server side: servers expose data sources and tools to any client that speaks the protocol
- Client side: applications connect to servers to reach that ecosystem without per-system integrations
- MCP apps: interactive apps run inside an AI client over the same protocol
- Documentation: core concepts, architecture, and build guides are published alongside the specification
Built with this
Integrated by
Agent Development Kit (ADK)
Google's open-source SDK for building, evaluating and deploying multi-agent systems.
Claude Agent SDK
SDK that exposes the Claude Code agent loop for building custom agents.
Cua
Open-source sandboxes, cross-OS desktop drivers, SDKs, MCP server and benchmarks for computer-use agents.
Dspy CLI
Been a DSPy fan for a while because it matched how I already think about LLM systems: not as prompts, but as programs yo