All Exercises¶

A quick-reference index of every hands-on exercise in the workshop. Click any exercise to jump to its page with full details.

Running exercises

You can run exercises individually from the terminal, or use the interactive Workshop TUI to browse and run them all from one place.

0 — Setup¶

Exercise	Description
Connection Test	Verify your OpenAI / Azure OpenAI connection is working

Exercise	Description
Chat Completion	Single-turn and multi-turn conversations with a travel assistant
System Prompts	Same query, different personas — see how system prompts shape behavior
Structured Outputs	Extract structured JSON from product reviews using Pydantic models

Concepts: Chat Completions API · System Prompts · Structured Outputs · What Is an Agent?

Exercise	Description
Function Calling	Define tools with `pydantic_function_tool()` and make a single-pass tool call
Tool Loop	Build the full agent loop: reason → call tool → observe → repeat

Concepts: Function Calling · The Agent Run Loop

Exercise	Description
Customer Support Agent	Multi-turn agent with order lookup, FAQ search, and refund processing

Pattern: Single Agent

Exercise	Description
Content Pipeline	Research → draft → edit pipeline with explicit context passing between stages

Pattern: Sequential

Exercise	Description
Stock Analysis	Fan-out/fan-in with 3 parallel analysts and an aggregator

Pattern: Concurrent

Exercise	Description
Brainstorm	PM, Designer, and Engineer debate a product idea in rounds
Maker-Checker	Code generator + reviewer in a reflection loop

Patterns: Brainstorm (Round-Robin) · Maker-Checker (Reflection)

Exercise	Description
Support Triage	Triage agent classifies queries and routes to billing, technical, or account specialists

Pattern: Handoff

Exercise	Description
Incident Response	Manager coordinates diagnostic, infrastructure, and communication workers

Pattern: Magentic