Model Context Protocol (MCP) Deep Dive: 2-Hour Interactive Workshop

certificate
Recording and certification are available to Multipass and Full ticket holders only
Please login if you have one.
Bookmark
Rate this content

Join a focused 2-hour session covering MCP's purpose, architecture, hands-on server implementation, and future directions. Designed for developers and system architects aiming to integrate contextual data with ML models effectively. 

Agenda:

- Introduction & Why MCP? Key challenges MCP solves and core benefits.

- Architecture Deep Dive: components, interactions, scalability principles. 

- Building Your Own MCP Server: guided walkthrough with code snippets and best practices; live demo or code review.

- Future of MCP Developments: potential enhancements, emerging trends, real-world scenarios.


Key Takeaways:

- Clear understanding of MCP's rationale.

- Insight into design patterns and scaling considerations.

- Practical steps to implement a prototype server.

- Awareness of upcoming trends and how to apply MCP in projects. 

This workshop has been presented at AI Coding Summit, check out the latest edition of this Tech Conference.

FAQ

MCP, or Model Context Protocol, is designed to allow large language models (LLMs) to communicate with various services without requiring developers to create an adapter for each service. This standardization makes it convenient for developers and end-users, similar to how USB type C works for devices.

MCP allows LLMs to access additional context and functionalities by providing a standard way to communicate with modern services. This helps LLMs offer more up-to-date and relevant information by interacting with external APIs and services efficiently.

The basic components of MCP include MCP host, MCP client, and MCP server. It also involves primitives such as tools, resources, and prompts. Communication between these components is bidirectional, using transport layers like stdio or HTTP and data layers like JSON RCP.

MCP's bidirectional communication includes layers like transport and data. The transport layer can use stdio or HTTP, while the data layer uses JSON RCP. This allows for efficient data exchange between the server and client, enabling features like sampling and elicitation.

Tools in MCP allow functions to be run remotely, resources can be easily cached and have different permissions, and prompts provide templates for LLMs to interact with an MCP server. These components help in organizing and managing interactions efficiently.

Discovery in MCP is aimed at making it a standard for any site or resource to provide definitions of their MCP servers. This feature, although not yet widely supported, will allow agents to automatically find and utilize MCP servers by accessing a well-known JSON file.

Currently, MCP primarily supports synchronous operations, but there are discussions about implementing asynchronous operations to improve its use in enterprise solutions where responses might take longer.

Future developments for MCP include server discovery, authorization on different levels, full open auth support, and SDK certification. There is also a push towards asynchronous operations and better session management to enhance scalability and reliability.

Developers can create an MCP server by installing the MCP SDK, setting up a server to handle requests via stdio or HTTP, and defining tools, resources, and prompts. Tools like MCP inspector can be used to debug and test MCP servers.

Just like USB type C standardizes the way devices connect, MCP standardizes the way LLMs communicate with various services. This eliminates the need for creating separate adapters for each service, similar to how USB type C simplifies connectivity across different devices.

Stepan Suvorov
Stepan Suvorov
86 min
15 Oct, 2025

Comments

Sign in or register to post your comment.
Video Summary and Transcription
Stepan Suvorov, MCTO at RoadSoft, shares insights on Model Context Protocol (MCP) at AI Coding Summit, highlighting its aim to enhance communication between LLM and services. MCP's internal structure includes host, client, server, tools, resources, and prompts, utilizing bidirectional communication with transport and data layers. Advanced features like discovery, sampling, and elicitation improve user-server interactions in MCP. Agent roles, efficient communication, MCP server creation, and tool registration are key aspects discussed, emphasizing community support and client setup for MCP development.
Video transcription and chapters available for users with access.

Watch more workshops on topic

AI on Demand: Serverless AI
DevOps.js Conf 2024DevOps.js Conf 2024
163 min
AI on Demand: Serverless AI
Top Content
Featured WorkshopFree
Nathan Disidore
Nathan Disidore
In this workshop, we discuss the merits of serverless architecture and how it can be applied to the AI space. We'll explore options around building serverless RAG applications for a more lambda-esque approach to AI. Next, we'll get hands on and build a sample CRUD app that allows you to store information and query it using an LLM with Workers AI, Vectorize, D1, and Cloudflare Workers.
High-performance Next.js
React Summit 2022React Summit 2022
50 min
High-performance Next.js
Workshop
Michele Riva
Michele Riva
Next.js is a compelling framework that makes many tasks effortless by providing many out-of-the-box solutions. But as soon as our app needs to scale, it is essential to maintain high performance without compromising maintenance and server costs. In this workshop, we will see how to analyze Next.js performances, resources usage, how to scale it, and how to make the right decisions while writing the application architecture.

Check out more articles and videos

We constantly think of articles and videos that might spark Git people interest / skill us up or help building a stellar career

Scaling Up with Remix and Micro Frontends
Remix Conf Europe 2022Remix Conf Europe 2022
23 min
Scaling Up with Remix and Micro Frontends
Top Content
This talk discusses the usage of Microfrontends in Remix and introduces the Tiny Frontend library. Kazoo, a used car buying platform, follows a domain-driven design approach and encountered issues with granular slicing. Tiny Frontend aims to solve the slicing problem and promotes type safety and compatibility of shared dependencies. The speaker demonstrates how Tiny Frontend works with server-side rendering and how Remix can consume and update components without redeploying the app. The talk also explores the usage of micro frontends and the future support for Webpack Module Federation in Remix.
Understanding React’s Fiber Architecture
React Advanced 2022React Advanced 2022
29 min
Understanding React’s Fiber Architecture
Top Content
This Talk explores React's internal jargon, specifically fiber, which is an internal unit of work for rendering and committing. Fibers facilitate efficient updates to elements and play a crucial role in the reconciliation process. The work loop, complete work, and commit phase are essential steps in the rendering process. Understanding React's internals can help with optimizing code and pull request reviews. React 18 introduces the work loop sync and async functions for concurrent features and prioritization. Fiber brings benefits like async rendering and the ability to discard work-in-progress trees, improving user experience.
Thinking Like an Architect
Node Congress 2025Node Congress 2025
31 min
Thinking Like an Architect
Top Content
In modern software development, architecture is more than just selecting the right tech stack; it involves decision-making, trade-offs, and considering the context of the business and organization. Understanding the problem space and focusing on users' needs are essential. Architectural flexibility is key, adapting the level of granularity and choosing between different approaches. Holistic thinking, long-term vision, and domain understanding are crucial for making better decisions. Effective communication, inclusion, and documentation are core skills for architects. Democratizing communication, prioritizing value, and embracing adaptive architectures are key to success.
Full Stack Components
Remix Conf Europe 2022Remix Conf Europe 2022
37 min
Full Stack Components
Top Content
RemixConf EU discussed full stack components and their benefits, such as marrying the backend and UI in the same file. The talk demonstrated the implementation of a combo box with search functionality using Remix and the Downshift library. It also highlighted the ease of creating resource routes in Remix and the importance of code organization and maintainability in full stack components. The speaker expressed gratitude towards the audience and discussed the future of Remix, including its acquisition by Shopify and the potential for collaboration with Hydrogen.
The Eternal Sunshine of the Zero Build Pipeline
React Finland 2021React Finland 2021
36 min
The Eternal Sunshine of the Zero Build Pipeline
For many years, we have migrated all our devtools to Node.js for the sake of simplicity: a common language (JS/TS), a large ecosystem (NPM), and a powerful engine. In the meantime, we moved a lot of computation tasks to the client-side thanks to PWA and JavaScript Hegemony.
So we made Webapps for years, developing with awesome reactive frameworks and bundling a lot of dependencies. We progressively moved from our simplicity to complex apps toolchains. We've become the new Java-like ecosystem. It sucks.
It's 2021, we've got a lot of new technologies to sustain our Users eXperience. It's time to have a break and rethink our tools rather than going faster and faster in the same direction. It's time to redesign the Developer eXperience. It's time for a bundle-free dev environment. It's time to embrace a new frontend building philosophy, still with our lovely JavaScript.
Introducing Snowpack, Vite, Astro, and other Bare Modules tools concepts!
Composition vs Configuration: How to Build Flexible, Resilient and Future-proof Components
React Summit 2022React Summit 2022
17 min
Composition vs Configuration: How to Build Flexible, Resilient and Future-proof Components
Top Content
Today's Talk discusses building flexible, resilient, and future-proof React components using composition and configuration approaches. The composition approach allows for flexibility without excessive conditional logic by using multiple components and passing props. The context API can be used for variant styling, allowing for appropriate styling and class specification. Adding variants and icons is made easy by consuming the variant context. The composition and configuration approaches can be combined for the best of both worlds.