from first concept to confident practice
Prompt Engineering
Write prompts that get reliable, checkable results - and prove it with tests.
About this track
Learn to write prompts the way professionals do: clear tasks with the reasons behind them, well-chosen examples, tags that separate instructions from data, long-context layouts, step-by-step reasoning, and structured outputs your code can trust.
Then treat prompts like software - templates under version control, evaluated on categorized test sets, compared version by version - and keep them safe from injection and data leaks. Along the way you’ll duel prompts, read an eval dashboard that hides a regression, fit documents into a context window and build a ticket-triage pipeline. The guidance follows current vendor best practices from Anthropic and OpenAI.
- No prior prompt engineering experience needed.
- Basic Python helps for the exercises.
- A modern browser with JavaScript enabled.
Your path
0 of 9 lessons · 0%Restoring your progress…
- Unit 1 · 0/3 lessons
Prompt foundations
Write clear requests, show examples, and structure prompts with tags, roles and templates. Badge: Clear communicator
- 0x001Write a clear promptState the task, the context and the reason behind it, the constraints and the output - specific enough that success is checkable. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
- 0x102Use examples to show the patternShow 3-5 diverse, consistent examples, balance the labels, and keep them clearly marked. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
- 0x203Structure prompts with tags, roles and templatesSeparate instructions from data with XML tags, use system and user messages well, and build reusable prompt templates. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
- Unit 2 · 0/4 lessons
Prompt design and testing
Lay out long context, chain and reason, request structured output, and test systematically. Badge: Prompt craftsperson
- 0x304Manage context and break down tasksLay out long documents for the best results, ground answers in quotes, and chain prompts for complex work. 22 min 0/2 exercises solved 22 min 0/2 exercises solved
- 0x405Let the model think: reasoning and verificationUse step-by-step reasoning where it helps, separate thinking from the answer, and check the steps. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
- 0x506Request structured, usable outputsDefine an output contract, use schema-constrained output where available, and parse and validate everything. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
- 0x607Test and refine prompts systematicallyBuild an eval set, grade with rubrics, compare versions per category, and catch regressions before shipping. 22 min 0/2 exercises solved 22 min 0/2 exercises solved
- Unit 3 · 0/1 lessons
Prompt safety
Understand injection and protect data with defenses outside the prompt. Badge: Boundary keeper
- Unit 4 · 0/1 lessons
Capstone
Render a triage prompt from config and evaluate a batch of replies. Badge: Prompt engineer