What is MCP?

MCP (Model Context Protocol) is a next-generation protocol designed to orchestrate and manage context between users, AI agents, and computational environments across decentralized systems.

Unlike traditional compute protocols focused purely on processing, MCP places emphasis on the context in which models (such as LLMs) and agents operate. This includes the user prompt history, the task objective, data access scope, verification rules, and the execution environment metadata — all bundled into a trusted, reusable context package.

MCP enables:

• Composable agent workflows where context travels across agents and models

• Secure execution by binding each task to a verifiable execution context (via zk or TEE)

• Decentralized coordination of model calls, API access, and storage use

• Modular architecture for integrating various model providers and runtime tools

— The ETMCP Core Team