AI Agents & MCP
Agents that call tools, use MCP servers, plan, and coordinate, and how to tell if they work.
Before this: AI Engineering.
-
12 min
-
15 min
Step 2How Anthropic builds effective agents
When to build an agent, and when a workflow is better.
How We Build Effective Agents: Barry Zhang, AnthropicAI Engineer
-
5 min
Step 3Tool calling
How a model decides to call your function.
What is Tool Calling? Connecting LLMs to Your DataIBM Technology
-
6 min
Step 4ReAct: reasoning plus acting
The pattern behind most agents.
ReAct AI Agents, clearly explained!Akshay Pachaar
-
34 min
Step 5Build an agent from scratch
A working agent in Python in about 30 minutes.
Build an AI Agent From Scratch in Python - Tutorial for BeginnersTech With Tim
microsoft/ai-agents-for-beginners★ 80.6k -
4 min
Step 6What is MCP?
The USB-C port for connecting AI to tools and data.
What is MCP? Integrate AI Agents with Databases & APIsIBM Technology
-
8 min
Step 7How MCP works
Hosts, clients, servers, and the messages between them.
How Model Context Protocol (MCP) actually worksGoogle Cloud Tech
-
26 min
Step 8Build an MCP server
Expose your own tools to Claude, Cursor, and others.
Build Anything With a CUSTOM MCP Server - Python TutorialTech With Tim
modelcontextprotocol/python-sdk★ 24.9k -
1h 13m
Step 9MCP, the full course
Servers, resources, prompts, and deployment.
Intro to MCP Servers – Model Context Protocol with Python CoursefreeCodeCamp.org
modelcontextprotocol/servers★ 92.1k -
10 min
Step 10Context engineering for agents
Four ways agents fail because of what's in their context.
Context Engineering for AI Agents: 4 Failure Modes to AvoidIBM Technology
-
16 min
Step 11Skills instead of agents
Package expertise as files an agent loads only when needed.
Don't Build Agents, Build Skills Instead – Barry Zhang & Mahesh Murag, AnthropicAI Engineer
-
18 min
Step 12Multi-agent systems
When splitting work across agents helps, and when it hurts.
Multi-agent Systems Explained in 17 MinutesShaw Talebi
openai/openai-agents-python★ 30.8k -
58 min
Step 13Agents with Hugging Face
Kick-off for a free, hands-on agents course.
Welcome To The Agents Course! Introduction to the Course and Q&AHugging Face
huggingface/agents-course★ 35.2k -
11 min
Step 14Evaluate agents
Measure trajectories and outcomes, not vibes.
How to evaluate agents in practiceGoogle Cloud Tech
SWE-agent/mini-swe-agent★ 9.9k
Build it
Watching isn't learning. Pick one and put it on GitHub.
- Write a 150-line agent with no framework: a model, three tools, a loop, and a log of every step.
- Turn something you use daily (notes folder, a database, a GitHub repo) into an MCP server and connect it to a desktop client.
- Write 20 tasks with checkable answers and measure your agent's success rate over 3 runs each.