Tool Use & MCP: Connecting AI to the World

By the end

What you'll build

• Explain why an LLM cannot act on the world or fetch live data on its own, and what a tool call adds.
• Write a valid tool definition in JSON Schema using type, properties, required, enum, and descriptions.
• Trace the think -> call -> observe -> repeat agent loop and identify its healthy stop conditions and step caps.
• Distinguish protocol-level errors (JSON-RPC error objects and codes) from tool-level errors (isError results) and craft each.
• Read a failing agent trace, locate the failure span, and classify it (e.g. ignored-error 'corrupt success').
• Describe MCP's host / client / server architecture, its transports, and the M x N problem it solves.
• Hand-craft the MCP initialize handshake and core JSON-RPC messages (tools/list, tools/call).
• Build a spec-shaped in-memory MCP server that dispatches initialize, tools/list, and tools/call with correct error codes.
• Run a real single-tool agent loop and reason about grading final state and required tool calls, not prose.
• Deploy a real MCP server on your own machine via the official quickstart and verify it via the initialize handshake and tools/list.

Curriculum

What's inside

6 modules · 34 lessons

1. 01

   Why LLMs Need Tools

   6 lessons
   • The Frozen Brain Problem
   • What a Tool Really Is
   • Core Vocabularyflashcard sr
   • Tools vs. Promptingquiz
   • The Request / Response Contract
   • + 1 more lesson
2. 02

   Tool Definitions with JSON Schema

   6 lessons
   • JSON Schema in 10 Minutes
   • JSON Schema Keywordsflashcard sr
   • Anatomy of a Tool Definitionfill blank
   • Validate Arguments Against a Schemapython lab
   • Descriptions Are Prompt Engineering
   • + 1 more lesson
3. 03

   Anatomy of the Agent Loop

   6 lessons
   • Think → Call → Observe → Repeat
   • Simulate the Looppython lab
   • Live Lab — Your First Real Tool Callagent lab
   • Loop States & Termsflashcard sr
   • Stop Conditions & Step Caps
   • + 1 more lesson
4. 04

   Structured Outputs & Error Handling

   6 lessons
   • Two Kinds of Errors
   • Build a JSON-RPC Error Responsepython lab
   • Error Analysis — Read a Failing Runtrace lab
   • Designing Tools That Fail Loudly
   • Live Lab — Make a Safe Writeagent lab
   • + 1 more lesson
5. 05

   MCP Architecture: Hosts, Clients, Servers

   5 lessons
   • Why MCP Exists: The M×N Problem
   • Hosts, Clients, Servers
   • MCP Primitivesflashcard sr
   • MCP Roles & Boundariesquiz
   • Craft the initialize Handshakepython lab
6. 06

   Capstone: Build a Mini In-Memory MCP Server

   5 lessons
   • The Capstone Plan
   • Server Part 1: initialize + tools/listpython lab
   • Server Part 2: tools/call + errorspython lab
   • Deploy a REAL MCP Server
   • Capstone Evidence Checkquiz

The shape of it

How this course works