AI Engineer’s Handbook to MCP Architecture

Author(s): Vatsal Saglani

Originally published on Towards AI.

Part 5: CLI, WebSocket, and Web UIBlog Header Generated by the Author

In our previous posts, we built a complete MCP architecture: specialized servers for file and command operations, a minimal MCP Client hub for routing, and an LLM layer that connects models to our tools. Now it’s time to make this stack accessible for us to play with.

In this blog, we’ll create two ways to interact with our system:

A rich command-line interface for local development and testingA WebSocket-based API for real-time streaming interactions integrated into a minimal web UI for quick manual testing

Each interface demonstrates different aspects of our architecture: the CLI shows the power of structured streaming output, the WebSocket API enables real-time interactions, and the integration of the WebSocket API to the UI brings it all together in a user-friendly package.

We’ll keep this blog a bit short and quick as it’s mostly about connecting the pieces we’ve already built. The focus is on showing how our tagged streaming format translates into real interfaces, whether that’s colorful terminal output or dynamic web updates. Think of it as the final layer that makes our MCP stack tangible and usable.

The complete code for this project is available on GitHub.

Contribute to vatsalsaglani/nano-MCP development by creating an account on… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI