Context-Aware AI: Hands-on with MCP (Model Context Protocol)

Workshop overview

Modern LLMs are powerful but operate in isolation without structured access to the tools, data, and systems developers rely on. This gap limits real-world usefulness, creates fragmented integrations, and increases security risks. The Model Context Protocol (MCP) addresses this by providing a standardized, interoperable, and secure way for AI models to connect with external resources.

Workshop Objectives

• To introduce participants to MCP and explain how it solves the integration and context-management challenges faced in AI-driven applications.
• To build a clear understanding of MCP’s architecture, components, and workflow patterns for real-world usage.
• To provide hands-on experience in creating an MCP server and connecting it with an AI client.
• To highlight best practices for secure, scalable, and maintainable MCP deployments.
• To help participants identify meaningful use-cases in their own domains where MCP can unlock new capabilities.

Note

• This workshop is of 3 hours duration.
• This is an in-person and hands on workshop.
• This workshop is beginner-friendly.
• Code & materials can be accessed at https://github.com/sdonapar/mcpdemo

Agenda

• Welcome & Introductions ( 15 mins)
  • Brief participant intro, workshop goals, agenda walkthrough. Set expectations: what attendees will learn by end.
• Overview: What is MCP & Why it matters ( 20 mins)
  • Definition of MCP.
  • Key motivation: LLMs are powerful but isolated from live data/tools; MCP solves “M×N problem” of integrations.
  • Architecture: client-server, tools/resources/prompts.
  • Real-world applications & ecosystem status.
• Core Concepts & Terminology ( 20 mins )
  • Walk through: servers, clients/hosts, tools, resources, prompts, workflows, context management.
  • Discuss advantages: standardization, interoperability.
  • Also highlight challenges: security, context bloat, governance.
• Break ( 15 mins )
• Hands-on Session: Build a Simple MCP Integration ( 40 mins)
  • Setup: choose a simple scenario (e.g., integrate an LLM with a data source or tool via MCP)
  • Installing SDK / creating MCP server/client.
  • Expose a simple data source (e.g., file system, database, or API) via MCP server.
  • MCP Inspector for debugging
• Deep Dive: Advanced Topics & Best Practices ( 30 mins)
  • Scaling & deployment of MCP servers (local vs remote)
  • Security / governance / access control (risks of tool-poisoning, context leakage)
  • Performance / context-bloat mitigation
  • Monitoring / observability
• Wrap-Up, Q&A ( 30 mins )
  • Summary of key take-aways
  • Provide additional resources
  • Open discussion

Pre-requisites

• Any Linux/Windows laptop
• install uv
• OpenAI API Keys/ Claude Desktop / Github Copilot
• Visual Studio Code

About the instructor

Sasidhar Donaparthi is working as Data Scientist working in a financial firm. Sasidhar has 25+ years of IT experience in manufacturing and financial services domains. He is a passionate Python programmer for a decade.

How to attend this workshop

This workshop is open for The Fifth Elephant annual members.

This workshop is open to 30 participants. Seats will be available on first-come-first-serve basis. 🎟️

Contact information ☎️

For inquiries about the workshop, contact +91-7676332020 or write to info@hasgeek.com