Free mini-course · Layer 1 of 5

Prompt Engineering: say exactly what you mean

The first layer. A prompt is the words a model is asked to act on — and most failures at every higher layer trace back to an instruction that was never clear in the first place. Nine lessons — the classic toolkit, then the frontier-model playbook: the effort dial, why-not-what prompting, and the fence.

promptscontextharnessloopsgraphs
LESSON 01

Instructions beat vibes

The single highest-leverage move in prompting is stating the task the way you'd brief a smart new colleague: what to do, what the output should look like, what to do when unsure. Vague asks produce confident guesses; explicit asks produce checkable work. Add the audience, the format, and the failure rule (“say I don't know rather than invent”) and half your quality problems disappear.

Do this now: Take a prompt you use often. Rewrite it with four explicit parts: role, task, output format, and uncertainty rule. Run both versions and compare.
LESSON 02

Show, don’t only tell: few-shot examples

Models pattern-match. One or three worked examples of input → ideal output teach format and judgment faster than paragraphs of description. Pick examples that cover an edge case, not just the happy path — the model learns from the differences between your examples.

Do this now: Add two examples to yesterday's rewritten prompt: one typical, one tricky edge case. Watch the edge-case handling change.
LESSON 03

Give reasoning room: chain of thought

For anything with steps — math, analysis, tradeoffs — asking the model to reason before answering measurably improves accuracy. Structure it: “First list the facts, then the options, then decide.” The trap: for trivial lookups, forced reasoning just burns tokens. Match the technique to the task.

Do this now: Ask the same hard question twice: once demanding an instant answer, once with “think step by step, then answer.” Compare where they diverge.
LESSON 04

Structure the output, or you can’t build on it

The moment a prompt's output feeds anything downstream — code, a spreadsheet, another prompt — free text becomes a bug. Demand a schema: JSON with named fields, a table, XML tags. This is the bridge to every higher layer: a node contract in a graph is just a structured-output prompt with enforcement.

Do this now: Take any extraction task and demand JSON with exactly three named fields. Then try to break it with messy input and tighten the prompt until it holds.
LESSON 05

Know the ceiling of this layer

Prompting cannot fix missing information, stale knowledge, or multi-step work that exceeds one context window. When you hit those walls you don't need a better prompt — you need the next layers: context (what the model sees), harness (what it can do), loops (how it iterates). Recognizing which wall you hit is the skill.

Do this now: Write down your last three frustrating model failures. Label each: wording problem, missing-information problem, or needs-tools-and-iteration problem. Only the first is a prompt problem.
LESSON 06

The effort dial: buy thinking, not words

Frontier models now expose an effort control — low, medium, high, xhigh — and it is the most misunderstood dial in prompting. Effort buys reasoning depth before the answer, not response length. Lowering it to get shorter replies gives you worse answers at the same length; raising it on easy tasks buys nothing but latency and cost. And the settings that were right for last year’s model are wrong for this one: newer models are so much stronger at low and medium that carried-over defaults quietly overpay.

Do this now: Take three tasks you run on high effort and re-run them at low and medium. Keep the cheapest setting whose output you cannot tell apart from high. Control length separately, with a one-line conciseness instruction.
LESSON 07

Tell it why, not just what

Old models needed step-by-step instructions; frontier models need context and goal — they work out the steps, usually better than you would specify them. The formula: “I’m working on [larger task] for [who it’s for]. They need [what the output enables]. Request: [one clear sentence]. Output format: [structure]. Constraints: [what must not happen].” The why-first prompt is shorter than the checklist it replaces, and the output is better — the model fills in the context you forgot to specify. This softens Lesson 02’s instinct: examples still teach format, but over-specified steps now cap the model at your imagination.

Do this now: Take your most instruction-heavy prompt and rewrite it as four parts: context-and-why, one-sentence request, output format, constraints. Delete every step-by-step instruction. Compare outputs.
LESSON 08

Delete your verification instructions

“Double-check your answer.” “Include a final verification step.” “Use a subagent to verify.” On older models these helped; on frontier models they stack on top of the self-verification the model already performs — wasted tokens, longer responses, no quality gain. Anthropic removed all of them from Claude Code’s own prompts and measured no loss and a significant cost drop. The deeper lesson: when a model misbehaves, most people add instructions. The engineers who built it removed instructions until the model could breathe. The guardrails were the problem.

Do this now: Search your saved prompts for “verify”, “double-check”, and “re-read”. Delete every instance, run your usual tasks, and watch whether quality actually moves. (Facts are the exception: frontier models are still confidently wrong sometimes — verify factual claims yourself.)
LESSON 09

Fence the agent: scope, narration, checkpoints

A proactive model will expand scope because it thinks it is helping. For agentic work the highest-value prompt is the fence: “Deliver what was asked, at the scope intended. If a better approach exists, say so in a sentence and continue with the task as asked.” Pair it with narration control — describe the communication style you want, positively, instead of listing don’ts — a subagent rule (“delegate only genuinely parallel work; don’t delegate what a few tool calls can finish”), and a checkpoint rule for autonomous runs: “pause only for destructive actions, real scope changes, or things only I can provide; otherwise keep going and report when done.”

Do this now: Add the scope fence and the checkpoint rule to one long-running agent task. Count the interruptions and the scope surprises against your last run without them.
Go deeper — the sources this course is built on:
Anthropic’s prompt engineering guide · promptingguide.ai (zero-shot, few-shot, CoT, self-consistency, ReAct and 15 more techniques) · OpenAI prompting guide · Anthropic’s official Claude Opus 5 prompting guide · @sairahul1’s “Claude Opus 5 Prompting Masterclass” — lessons 06–09 synthesize its plain-English playbook: the effort dial, why-not-what, deleting verification instructions, and the agentic fence