The Model Context Protocol (MCP)
by Bytetality • October 10, 2026
Learn about the Model Context Protocol (MCP), a new open protocol from Anthropic that standardizes how applications provide knowledge and capabilities to Large Language Models (LLMs). Explore its features, benefits, and implementation.
The rise of Large Language Models (LLMs) like Claude has revolutionized how we interact with AI. However, a key limitation has been the difficulty in seamlessly integrating these models with external data sources and tools. Traditionally, developers faced a fragmented landscape, requiring custom integration code for each data source and tool, limiting LLM capabilities to information already present within the model’s training data.
Enter the Model Context Protocol (MCP), an open protocol developed by Anthropic designed to solve this problem. This article will provide a comprehensive overview of MCP, its architecture, and how it’s changing the way developers build AI-powered applications.
Overview
The MCP protocol is fundamentally about standardization. It establishes a universal way for applications to provide context – data, tools, and capabilities – to LLMs. Think of it as a common language that allows diverse systems to communicate effectively with an LLM, regardless of the underlying model.
Created by Anthropic, the company behind Claude, MCP aims to simplify the integration process, reduce development complexity, and unlock new possibilities for AI assistants.
MCP isn’t just a protocol; it’s a well-defined architecture built around three core primitives:
Tools
These are executable functions that an LLM can invoke to perform specific tasks – from querying databases to making API calls. They represent the actions the LLM can take.
Resources
These are data sources – like files, databases, or APIs – that the LLM can access for information. They’re the “knowledge” the LLM needs.
Prompts
These are reusable templates used to guide the LLM's responses, providing context and structure to conversations. Beyond these core primitives, MCP supports:
- Sampling: Controlling the diversity of LLM outputs.
- Elicitation: Requesting specific information from the LLM through structured prompts.
- Roots: Providing access to client-side root directories or contexts.
Notifications
Enabling real-time updates and status reporting between the client and server. The protocol leverages JSON-RPC 2.0 as its message format, ensuring a lightweight and language-agnostic communication layer. MCP supports both STDIO (a traditional terminal-based interface) and HTTP transports (including SSE and Streamable variants), offering flexibility for different deployment scenarios.
Performance & Practical Use
The Quarkus MCP server implementation, built on the Quarkus framework, provides a robust and efficient solution for integrating with LLMs. The protocol is designed for bidirectional communication, allowing both the client and server to exchange information seamlessly. This enables features like:
Client-to-Server Requests: Clients can invoke server-side tools or access resources.
To learn more about MCP, see the following:
