Hands-On: Build Your Own Agent

Time to get serious -- you're building your own agent!

Over the past five modules, you've laid the theoretical groundwork: You understand how LLMs work under the hood, you've mastered advanced prompt engineering, you know AI coding tools inside and out, you've internalized agentic AI patterns, and you understand how MCP integrations work. Now you're bringing it all together.

In this module, you'll build a working agent step by step -- from the system prompt to tool integration to cost optimization. By the end, you won't just understand how agents work -- you'll have built and run one yourself.

*Learning Objective

After this module, you'll be able to set up an agent with a system prompt, tool integration, and MCP connectivity, analyze its ReAct loop, and calculate its running costs. You'll be operating at Bloom's levels of "Apply" and "Analyze."

Prerequisites

Before you start, make sure you've completed the following modules:

Module 01 -- LLM Deep Dive: You need the understanding of tokens, context windows, and model differences
Module 02 -- Prompt Engineering: System prompts and chain-of-thought are the foundation for agent instructions
Module 03 -- AI Coding Tools: You know the tool landscape and understand agent-driven development
Module 04 -- Agentic AI Patterns: ReAct, tool use, and plan-and-execute are the architecture of your agent
Module 05 -- MCP Integrations: You know how MCP servers are built and connected

!Technical Requirements

For the hands-on exercises, you'll need: Python 3.10+ or Node.js 18+, an API key for Claude or GPT, and a text editor. Code examples are shown in Python and TypeScript -- you can choose whichever language you prefer.

What to Expect

This module consists of four building blocks, each one building on the last:

Step 1 -- Agent Setup: You define the system prompt, choose a model, and write your first tool definition. By the end, you'll have an agent that responds to instructions.

Step 2 -- Connect an MCP Server: You connect your agent to an existing MCP server and test whether the communication works. Your agent gains its first real capabilities.

Step 3 -- Understand the Agent Loop: You observe your agent in action -- how it thinks, acts, and observes. You walk through predefined scenarios and analyze the behavior.

Step 4 -- Calculate Costs: You calculate what your agent costs per day and per month, and optimize it for a realistic budget.

Ready?

Let's get started -- with the setup of your first agent.

Reflect

In this hands-on module, you will build your first own agent -- from setup to MCP integration to cost calculation. Each step builds on the previous one. Let us start with the setup of your first agent.