The Art of the Prompt

A free, interactive masterclass · 2026 edition

Mastering the
Prompt.

A course on directing large language models, from your first prompt to agentic, reasoning-model, evaluation-driven mastery. Built to be done, not just read: every module carries labs, a rubric, and prompts you can run as you go.

Orientation · Course contract

A prompt is a thinking instrument, not a magic spell

Most people treat prompting as guessing the secret words. It isn't. A prompt is an engineered brief you hand to a capable but context-free collaborator. This is a course, not a cheat sheet: each module states what you'll be able to do, gives you something to practice, and shows you how to judge your own work.

The one idea to anchor everything

Treat the model like a brilliant new colleague on their first day: phenomenally knowledgeable, eager, fast, but with zero knowledge of your goals, your norms, or what "good" looks like to you. Every technique here is a better way to brief that colleague. The golden test: if you showed your prompt to a real person with no extra context and they'd be confused, the model will be too.

How to use this course

Examples tuned for

The skill is identical for everyone; this only swaps the worked examples toward your world. Change it anytime, we'll remember your choice. See the full business prompt collection →

Accessibility & inclusion note

This page is built to WCAG-minded standards: semantic headings, a skip link, visible focus styles, text alternatives on diagrams, AA-contrast text, and full support for prefers-reduced-motion. Examples deliberately span many roles, domains, and one multilingual and one fairness-sensitive case. If you adapt this into slides or video, add captions, transcripts, and spoken descriptions of every diagram from the start rather than retrofitting later.

Backward design

What you'll be able to do

This course is designed outcomes-first: every lab and rubric maps back to these. By the end, you will be able to:

Course-level learning outcomes

Why the structure changed in this edition

This revision teaches the modern reasoning-model default first (the reasoning-model module), then the historical techniques as context (the advanced-reasoning module). That ordering prevents a common trap: learning 2023-era scaffolding as if it were still the default, then having to unlearn it. You'll test claims against real models rather than memorising them.

Prerequisites

Pick your track

"beginner to advanced" only works if you know which path is yours. Everyone does the Core modules; the rest depends on your track. No coding is required for two of the three. Pick one to tailor the course map below.

A

No-code user

For: anyone using ChatGPT, Claude, or Gemini in a chat box.

Prereqs: none.

Do: Modules 1–4, 6, 8. Skim 5, 7, 9.

B

Power user / knowledge worker

For: writers, analysts, marketers, ops, educators building repeatable workflows.

Prereqs: comfort with structured documents; JSON helpful, not required.

Do: all modules; labs in 3, 5, 7.

C

Developer / builder

For: engineers wiring prompts into apps, agents, and pipelines.

Prereqs: APIs, JSON Schema, basic tool-use concepts.

Do: all modules + every lab + capstone.

The full course map

0
Foundation, stay in the driver's seat
The LLM as a tool for business, learning & life, not a crutch.
Start here≈ 20 min
1
Foundations, the anatomy of a prompt
Five components, clarity, the "why", positioning.
Core≈ 45 min
2
Core techniques & output control
Zero/few-shot, roles, delimiters, structured output, templates.
Core≈ 60 min
3
Modern reasoning-model prompting
The 2026 default mental model. Goal + constraints + contract.
Core≈ 60 min
4
Advanced reasoning, historical patterns
CoT, self-consistency, ReAct, ToT, and when NOT to use them.
Advanced≈ 50 min
5
Context engineering & agentic prompting
System prompts, the five inputs, tools, autonomy.
Advanced≈ 75 min
6
Anti-patterns & debugging
Nine failure modes + a systematic diagnosis flow.
Core≈ 45 min
7
Evaluation, rubrics & iteration
Build a test set and grader; iterate without regressing.
Advanced≈ 75 min
8
Safety, ethics & responsible prompting
Bias, privacy, prompt injection, provenance, oversight.
Core≈ 50 min
9
Provider portability & tuning
Claude vs GPT/o-series vs Gemini, what changes, what stays.
Optional≈ 40 min
Capstone, prompt portfolio + eval suite
Ship and defend a real, evaluated, provider-portable prompt.
All tracks≈ 90 min
Why most tutorials are already stale

What changed in 2025–2026

Before any module, hold these four shifts in mind. The discipline split in two: models now reason on their own, and several classic "tricks" are now redundant, or counterproductive, on frontier models.

01

Many frontier models reason natively

Flagship models (Claude Opus 4.x, OpenAI o-series / GPT-5, Gemini 2.5/3 with thinking) can "think" before answering. On those models, hand-written scaffolds like "think step by step" and elaborate few-shot examples are often unnecessary. The skill shifted from telling the model how to think to telling it what to achieve, but always test on the model you deploy.

02

Prompting became context engineering

The real job is curating the useful set of tokens in the window, system prompt, history, tool results, retrieved knowledge, not wording one perfect sentence.

03

Structure can be enforced

Constrained decoding / structured outputs let you request schema-valid JSON with strong guarantees instead of pleading "please return valid JSON." This replaced a genre of fragile prompt hacks.

04

Knobs supplement scaffolds

Instead of writing reasoning steps by hand, you can turn a dial: an effort / thinking control. The advanced move in 2026 is frequently to remove instructions and let the model's own reasoning run at the right intensity.

Counter-intuitive headline for 2026

Advanced does not always mean more elaborate. On frontier reasoning models, the expert move is often to strip a prompt down, fewer examples, no "you MUST", no manual chain-of-thought, and instead give a clean goal, hard constraints, an output contract, and the right effort setting. Verify the effect on your target model rather than assuming it.

From the makers · put it to work

Done learning? Now make AI earn its keep.

This manual is free because it is published by people who do this for a living. The course taught you to direct AI. These two put it to work, one for your business, one for your plan.

REF · A-01 Ascend Creative Consulting Premium websites and practical AI systems that turn attention into booked calls, quotes, and better leads. Human-AI integration, not random AI. Visit Ascend → REF · P-01 PlanMason A coached business-plan builder for founders who need a yes from the people paid to say no. Built, not generated. Try PlanMason →
Keep this open

The master cheat sheet

Universal prompt skeleton

You are a {{role}}. Task: {{the single clear objective}} Context: - Audience: {{who reads this}} - Why it matters: {{motivation}} - Source material: {{wrapped in tags}} Constraints: - {{hard limits, length, scope, tone}} - Do X (state positively, not "don't do Y") Output format: - {{exact shape, prose / JSON schema / sections}} Before finishing, verify the result against: {{named criteria}}

Decision quick-reference

  • Simple task → zero-shot, Task + Format only.
  • Specific format/tone → add a few clean examples (fewer on reasoning models).
  • Hard reasoning, capable model → goal + constraints, raise effort, test before adding CoT.
  • Must be correct → named verification step (and self-consistency if critical).
  • Output feeds a system → structured output / schema.
  • Tools / multi-step → lean system prompt, autonomy default, safety rails.
  • Untrusted content → injection-defence wrapper (the Safety module).

When a prompt fails, in order

  1. Is the task unambiguous? Add the missing component.
  2. Did you give the why?
  3. Is the key instruction first or last, not buried?
  4. On a reasoner, try removing scaffolding, not adding it.
  5. Still failing? Iterate against a test set, one change at a time.

Three reusable templates, copy or launch

Minimal reasoning-model
Task: [single objective]

Constraints:
- [scope, tone, exclusions]
- [success criteria]

Output:
[exact sections / schema]

Before finishing, check the result against [named criteria].
Retrieval-grounded
<documents>
[paste your source text here]
</documents>

Question: [your question]

1. Extract the quotes relevant to the question.
2. Answer using only those quotes.
3. Cite each source inline.
4. If the evidence is thin or missing, say so plainly.
Prompt-evaluation
Grade the prompt below from 1-4 on: objective clarity, context sufficiency, output contract, robustness to ambiguity, and safety + evidence.

Return JSON: {"scores": {...}, "highest_risk": "...", "one_best_improvement": "..."}

Prompt to grade:
"""
[paste the prompt you want graded]
"""
Reference

Glossary

Acronyms are defined on first use in the modules; this is the quick reference.

Zero-shotPrompting with instructions only, no examples.
Few-shotProviding example input→output pairs to steer behavior.
CoT (Chain-of-Thought)Eliciting step-by-step reasoning before the final answer.
Self-consistencySampling multiple reasoning paths; taking the majority answer.
ReActReason→Act→Observe loop; the basis of tool-using agents.
ToT (Tree-of-Thoughts)Branching, scoring and pruning multiple reasoning paths.
Context engineeringCurating the useful set of tokens across a task.
Structured outputsConstrained decoding that yields schema-valid output.
Effort / thinking budgetA control for how much a model reasons before answering.
System / developer messageThe authoritative instruction defining role and rules.
RAG (Retrieval-Augmented Generation)Grounding answers in fetched documents.
Prompt injectionUntrusted content trying to hijack the model's instructions.
LLM-as-judgeUsing a model to score outputs against criteria at scale.
Common questions, answered

Prompting FAQ

Short, direct answers to the questions people ask most about prompting and AI models.

What is prompt engineering?
Prompt engineering is the practice of writing clear, specific instructions that get a large language model to do what you want. A strong prompt names the task, gives the model the context it needs, and specifies the output format. It is iterative: you test, see where the model misreads, and refine.
How do you write a good prompt?
Use five components: a role, a clear task, the relevant context, the output format, and any constraints. State what you want the model to do rather than what to avoid, give the reason behind important rules, and put your key instruction at the start or end of the prompt, where the model attends most.
What are the components of a prompt?
Role (who the model should act as), Task (the single objective, always required), Context (audience, background, source material), Format (the shape of the output), and Constraints (limits, tone, length). Start with the Task and add the others only when the result misses.
What is the difference between zero-shot and few-shot prompting?
Zero-shot prompting gives only instructions and no examples; few-shot prompting includes a handful of input-output examples to lock in a format or behaviour. Try zero-shot first, especially on reasoning models, and add 2 to 5 clean, varied examples only when you need to pin down a specific style or edge case.
How do you prompt reasoning models differently?
Reasoning models think before answering, so older scaffolding can hurt them. Give a clear goal, hard constraints, and an output contract, then let the model reason. Skip "think step by step" and heavy few-shot examples; if reasoning is the gap, raise the effort or thinking setting instead of rewriting the prompt.
How do I stop an AI from making things up (hallucinating)?
Ground the model in sources, ask it to cite the evidence for each claim, and tell it to say "I don't know" when the information is not there. Add a verification step with named criteria, and for anything high-stakes, check the facts rather than trusting fluent prose.
What is chain-of-thought prompting?
Chain-of-thought prompting asks the model to reason step by step before giving a final answer, which historically improved math and logic tasks. On modern reasoning models it is often unnecessary or even counterproductive, since they already reason internally, so test whether a simpler prompt does better first.
Which AI model should I use for which task?
Use a fast, inexpensive model for simple, well-defined work like classification, extraction, and formatting, and reserve the large reasoning model for ambiguous, multi-step, or high-stakes problems like analysis, planning, and complex code. Running the heaviest model on light tasks is the fastest way to burn through your rate limits and budget.
What is a system prompt?
A system prompt is the top-level instruction that defines the model's role, the rules it must follow, and the deliverable for a whole task or application. Keep it lean: front-loading every possible rule forces the model to carry all of it even on trivial requests.
What is RAG (retrieval-augmented generation)?
RAG grounds a model's answer in documents you retrieve and place into the prompt, so it can use current or private information beyond its training data. Wrap each source in tags with its origin, ask the model to extract the relevant quotes first and answer only from those, and require inline citations.
What is a context window?
The context window is the total amount of text, measured in tokens, that a model can consider at once, including your prompt, any documents, the conversation history, and its own answer. Models attend most to the start and end of the window, so place critical instructions there rather than burying them in the middle.
How do I get better results from ChatGPT, Claude, or Gemini?
Be specific, give context, and say what you want explicitly. Ask the model to interview you for any missing context before it answers, request the exact output shape, and iterate rather than expecting a perfect first try. Treat it like a brilliant new colleague who has zero knowledge of your goals.
Free download

Get the 50-prompt pack

Drop your email and I'll send the full run-it-now prompt library as a Notion template + PDF, plus a fresh batch of prompts each month. No spam, unsubscribe anytime.

You'll also get a heads-up when new modules drop.