Building Agentic AI with Model Context Protocol 

Duration: 2 Days / 16 hrs

Introduction 

This two-day, hands-on course teaches how to build agentic AI systems using the Model Context Protocol (MCP). Participants learn how to structure context, expose data and tools, enable RAG, and design secure, scalable agent workflows that go beyond simple prompt-based AI. 

Objectives 

By the end of this course, participants will be able to: 

  • Understand MCP architecture and its role in agentic systems 
  • Expose data, prompts, and tools through MCP servers 
  • Enable RAG, tool-use, and sampling with human-in-the-loop control 
  • Design and build basic agentic AI applications using MCP 

Key Takeaways 

Participants will leave with: 

  • A clear mental model of MCP and agent-based AI design 
  • Practical experience building MCP-powered AI workflows 
  • The ability to design secure, extensible, real-world agentic applications 

Course Outline: 

1. Introduction to Model Context Protocol 

  • Why MCP? 
  • Core architecture and components 
  • Hosts, models, and contexts 
  • MCP layers, protocols, and connections 
  • Message types and purposes 

2. Enabling RAG with Resources and Roots 

  • Exposing data with resources 
  • Building resource URIs and types 
  • Reading resources from the server 
  • Resource discovery with templates 
  • Updating resources in real-time 
  • Establishing focus with roots 

3. Building Patterns with Prompt Templates 

  • Prompt structure and discoverability 
  • Multi-step workflows 
  • Embedding resources in prompts 
  • Application integration 

4. Enabling Tool-Use on Servers 

  • Defining tools for LLM readability 
  • Patterns of tool use 
  • Tool discovery and validation 
  • Error and exception handling 

5. Using LLMs from MCP Servers with Sampling 

  • Sampling mechanics and purpose 
  • Writing sampling requests and responses 
  • Human-in-the-loop integration 

6. Security and Best Practices 

  • Secure design principles 
  • Robust and extensible architecture 

7. Building Agentic Applications 

  • Agent placement and lifecycle 
  • Integration of LLMs, tools, and resources 
  • Simple agentic application development 
  • Testing and debugging 
  • Future directions