← All 45 books AI prompting techniques, one technique per page Get the full edition · £10
One technique per page

AI prompting techniques, one technique per page

The twenty-five asks that take you from 'why is it ignoring me' to answers you can use.


Steve Hodgkiss 8 techniques

A diagram of the mechanism, the classic mistake, and one thing to go try. That's a page.

AI prompting techniques, one technique per page

The twenty-five asks that take you from 'why is it ignoring me' to answers you can use.


Set in Space Grotesk, Inter and JetBrains Mono (SIL Open Font License).

Techniques checked against the official prompting guides: OpenAI's prompt engineering guide (platform.openai.com/docs/guides/prompt-engineering), Anthropic's prompt engineering documentation (docs.anthropic.com), and Google's Gemini prompt design strategies (ai.google.dev). The name 'chain of thought' is cited historically from Wei et al., Chain-of-Thought Prompting Elicits Reasoning in Large Language Models, arXiv:2201.11903 (2022). No benchmark numbers are quoted; every technique is presented with the reasoning behind it, not as magic.

General craft guidance only. The assistant can be wrong; verify anything that matters.

Your purchase is for personal use only. You do not have redistribution rights: please do not share, resell, or republish this book or its pages.

© 2026 Steve Hodgkiss. All rights reserved. Personal use only; no redistribution rights.

Edition 1.0 · stevehodgkiss.net

Contents

Contents


Part 1 · The conversation4
The context window is a desk5
System message and user message6
Part 2 · Getting answers7
The single clear ask8
Show a filled example: few-shot9
Constraints up front: format, length, audience10
Part 3 · Better thinking11
Show your working: chain of thought12
Iterate: the first answer is draft one13
Part 4 · Working with output
Verify facts: trust but check14
Part 1 of 4
What you're actually talking to
1

The conversation

Before any technique: what the assistant sees when you type. Five pages on the context window as a desk, the two kinds of message, why it only knows what you give it, what a fresh chat forgets, and what really happens when the context overflows.


In this part
  1. 01The context window is a desk
  2. 02System message and user message

The context window is everything the assistant can see of your conversation: your messages, its replies, your pasted documents. It reads all of it, and only it, every turn.

Prompting · No. 01
The conversation

The context window is a desk

It only sees what's on it

THE CONTEXT WINDOW IS A DESK: IT HOLDS ONLY WHAT YOU'VE PUT ON IT.your messagesassistant repliesyour pasted docsthe context windowyou addevery reply, it re-reads the whole desk: nothing outside it exists

Let's say you imagine it remembers your whole relationship. It reads one surface: the context window, every turn.

The mistake is treating it like an old colleague who was there for all of it.

Anthropic's docs picture it as a brilliant new employee: it knows only this conversation.

It only knows what's on the desk.

TRY IT THIS WEEK

Open your longest active chat and scroll up until you find a rule you stated early that the newest replies have drifted from. That drift is the desk.

A system message (custom instructions, in chat tools) is read before every turn and carries more authority than an ordinary message. Your everyday asks sit below it, in the flow.

Prompting · No. 02
The conversation

System message and user message

Rules pinned above the chat

TWO KINDS OF MESSAGE: PINNED RULES THAT OUTRANK, AND THE CHAT THAT FLOWS.terse - plain Englishno apologiessystem message:standing rules,pinned aboveuser: tighten this summaryassistant: here it is, 3 linesread beforeevery turnthe pinned note is read every turn: the bubbles come and go

Let's say your rules get ignored three messages in. Pin them in the system message: read every turn, and it outranks the chat.

Buried in an ordinary message, a rule competes with everything after it.

Every major chat tool has the box: custom instructions, or the system field.

Rules pinned above the desk, not left lying on it.

TRY IT THIS WEEK

Find your tool's custom-instructions box and pin your three standing rules: tone, spelling, and the one thing it always gets wrong.

Part 2 of 4
The ask itself
2

Getting answers

The five moves that fix most mediocre answers. One clear ask, a role, a filled example, constraints stated up front, and options instead of a single take.


In this part
  1. 01The single clear ask
  2. 02Show a filled example: few-shot
  3. 03Constraints up front: format, length, audience

One ask carries one verb, one target and one shape. Vague asks get an average of everything the assistant guesses you might have meant.

Prompting · No. 03
Getting answers

The single clear ask

One verb, one target, one shape

ONE ASK, ONE TARGET. VAGUENESS GETS YOU AN AVERAGE OF EVERYTHING.weak ask"help with this"revise?rewrite?summarise?proofread?clear ask"cut to 3 bullets"one verb, one target, one shapecarries:the verbthe materialthe shapethe deadlinesay what to do, not what not to do: the guides all lead with that

Let's say you type "help with this" and get mush back. The assistant averaged every reading of your ask.

"Help with this" could mean four jobs. It picked one at random.

"Cut the attached to 3 bullets for my manager" has verb, material, shape, reader.

One ask, one verb, one shape.

TRY IT THIS WEEK

Take today's mushiest ask and rewrite it as verb plus material plus shape plus reader. Send both versions, keep the better one.

A filled example (few-shot) shows the exact pattern: the fields, the labels, the casing. The assistant copies the shape it saw far more reliably than a shape you described.

Prompting · No. 04
Getting answers

Show a filled example: few-shot

Filled sample forms teach the pattern

TWO FILLED SAMPLE FORMS TEACH THE PATTERN. THE THIRD FILLS ITSELF.sample forms you filledin: bug reportout: PRIORITY HIGHin: typoout: PRIORITY LOWin: crashout: PRIORITY HIGHsame patternit copies theSHAPE it saw,not a shapeyou describedOpenAI's guide: a handful of diverse examples beats a paragraph of description

Let's say you describe a format and still don't get it. Show one filled example instead: in, out, exactly as you want it.

Describing a format in prose leaves the details, casing and column order, to guesswork.

Two or three examples, varied, beat a paragraph. All three vendors say so.

Don't describe the form. Hand over a filled one.

TRY IT THIS WEEK

Take any recurring task and write two filled examples of input and output, then hand the assistant a third input. Keep the examples in a note for reuse.

Format, length and audience are boundaries, set before the work starts. Stated up front, the first draft arrives inside the lines.

Prompting · No. 05
Getting answers

Constraints up front: format, length, audience

Draw the lines before the work

STATE THE BOUNDARIES BEFORE THE WORK: FORMAT, LENGTH, AUDIENCE.format: tablelength: 5 rowsreader: my boardthe ask| option | cost || a | low || b | mid |arrives inside the lineswithout gates: essay20 bulletsfor nobodyput constraints before the task, not as an apology after the first draft

Let's say every answer needs the same trim afterwards. You set the boundaries after; set them before.

Format guessed, length open-ended, reader imagined: that's three rewrites you do yourself.

One line: table, five rows, for the board. The first draft fits.

Boundaries before work, not apologies after.

TRY IT THIS WEEK

Add a standing constraint line to your three most frequent asks: exact format, hard length limit, named reader. Watch the trim work disappear.

Part 3 of 4
Reasoning, out loud
3

Better thinking

Answers get better when the thinking shows. Five pages on asking for the working, self-critique, iteration, decomposition, and defining good before you ask.


In this part
  1. 01Show your working: chain of thought
  2. 02Iterate: the first answer is draft one

Ask for the steps before the conclusion. Visible working turns a wrong answer into a findable slip, and a right one into a checkable one.

Prompting · No. 06
Better thinking

Show your working: chain of thought

Checkable steps, not a leap

WORK ON THE PAGE BEFORE THE ANSWER: YOU CAN CHECK WHAT YOU CAN SEE.answer only= 41wrong, and youcan't see whereworking first, answer last1. list the fees: 12 + 92. 3% of 400 = 123. 21 * 2% ... carry= 47spot the slip in step 3,not in the totalthe name is historical, Wei et al. 2022; every major model now has a thinking mode you can switch on

Let's say it answers 41 and you can't tell if that's right. Ask for the working before the answer: steps, then total.

A bare number you must trust is a number you cannot correct.

Named in a 2022 paper, now built in: your tool likely has a thinking mode. For big sums, still say "show the steps".

Work on the page, where you can check it.

TRY IT THIS WEEK

Take one real estimate you'd normally accept blind and ask for it twice: once bare, once with numbered steps shown before the answer. Compare.

The first answer is draft one, not the verdict. Each revision of the ask, informed by reading the draft, climbs the same chat.

Prompting · No. 07
Better thinking

Iterate: the first answer is draft one

Prompting is a loop

PROMPTING IS A LOOP: DRAFT, READING, REVISION OF THE ASK. NOT A COIN TOSS.draft oneread it criticallybetter askagain, with what you learnedeach pass costs one message: cheaper than doing the work twice

Let's say one shot misses and you conclude the tool can't do it. It rarely can, first pass: read the miss, revise the ask.

Retyping the same ask louder, or giving up after one try.

Second asks are cheap: "shorter, second person, lead with the deadline".

The first answer is draft one.

TRY IT THIS WEEK

Take one answer you were disappointed by this week and give it two honest revision passes instead. Note what each pass actually fixed.

The assistant can produce a fluent sentence about something that is not true. Verify what carries weight: numbers, names, quotes, anything with consequences.

Prompting · No. 08
Working with output

Verify facts: trust but check

Confident is not the same as correct

CONFIDENT AND CHECKED READ IDENTICALLY. VERIFY WHAT CARRIES WEIGHT."revenue rose 34% in Q2"reads the same either wayshipped as-is,it was inventedchecked at thesource, correctedwhat checking looks like1. ask for a source for every number2. open the ones that carry weight3. paste text in, ask from it onlythe stakes rule: how bad is itif this number is wrong?the docs are blunt: models can fabricate, and reference text is the counter-weight

Let's say a number in its draft looked right, so it stayed. Fluency is free; correctness isn't. Check the load-bearing lines.

Reading it over and calling that verification: plausibility is not proof.

Paste the source in, ask "from this text only", and request a source per claim.

Trust the draft. Check the facts.

TRY IT THIS WEEK

Take one assistant-produced number going into real work this week. Trace it to a source you'd cite. If you can't, cut it or fix it.

Index

Index


Constraints up front: format, length, audience10
Iterate: the first answer is draft one13
Show a filled example: few-shot9
Show your working: chain of thought12
System message and user message6
The context window is a desk5
The single clear ask8
Verify facts: trust but check14