hotpromptThe framework

The framework

Everything hotprompt knows about what makes a prompt work, in full. 11 moves that apply to any ask, then 21 kinds of ask with their own 133 moves and 133 things worth leaving out. This page is generated from the same file the engine reads, so it cannot describe a version we don’t run.

How to read it

A move is one decision a good prompt makes on purpose. When you sharpen something, every move gets a mark: ✓ if your ask already made that decision, + if we filled it with a sensible default, _ if it is a thing only you could know — the facts, the source text, the error message. Those last ones are left blank deliberately. A model that invents them is guessing, and a guess is the failure mode this whole thing exists to prevent.

The defaults printed below are what the + marks actually insert. They are starting points chosen to be right more often than not, not laws — override any of them in the second box.

The 11 universal moves

These apply whatever you are asking for. A domain below may rename one or sharpen it, but it is answering the same question.

Audience & relationship
Who the output is for — their identity, expertise level, and your relationship to them; the single biggest driver of register and framing.
Goal / desired outcome
The one result the output should produce — a single concrete action or decision, not a menu of asks.
Context & source material
The real facts, existing text, data, or situation to work from — supplied by the user, never invented by the model.
Tone & voice
The register and personality of the output, stated concretely and paired with what to avoid.
Format, length & shape
The container: exact artifact, length cap, and structure so the result drops in without rework.
Constraints & must-avoids
Hard filters and exclusions — what to leave out, steer clear of, or never do.
Referent disambiguation
Resolve or flag a polysemous subject term (transformers, python, swift, mercury) before sharpening — pick the most likely reading and state the assumption, or ask.
Source-material gate
For rewrite / summarize / edit / translate tasks, require the actual text or data up front; if the ask says 'this' or 'it' with nothing attached, ask for the source rather than inventing one.
Request polarity
Capture the DIRECTION of a requested change (less X, more Y, 'funny but NOT a roast') relative to the source, instead of snapping to a fixed default register.
Subject-shape branch
Before applying domain moves, detect figure-vs-environment, function-vs-pattern, prose-vs-fixed-form, so person- or function-centric defaults don't fire on the wrong shape.
Clarify vs fill
For genuinely underspecified asks, prefer one targeted clarifying question over silently filling every default.
The 21 domains

Email & outreach

Routed here by email, cold email, reach out, follow up, follow-up, outreach, intro, reply to +11 more

  1. +Recipient & relationshipAssume a cold recipient with no prior relationship; open by placing yourself (who you are + why them specifically) in one line.
  2. +Goal / the askPick one concrete, easy-to-say-yes-to ask and make it explicit in a closing line.
  3. +Why-you-why-them hookLead with one specific, true reason tied to the recipient, not a generic compliment.
  4. +Tone & registerWarm, professional, direct; conversational not chummy; no corporate filler or hype.
  5. +Length & shapeUnder 120 words, 3 short paragraphs, one clear ask, plus a subject line.
  6. +Subject lineOffer 2-3 options: specific, under 6 words, no clickbait or all-caps.
  7. +Call to action / next stepClose with a specific, low-effort next step actionable in a single reply.

Left out for this type: role prompting ('world-class copywriter'); generic 'clear and well-structured' format; SEO keywords, hashtags, or emoji; long background dumps / 'include everything'; multiple asks or a menu of options; fake over-personalization theater; urgency/scarcity/flattery persuasion stuffing

Code & engineering

Routed here by write a function, write a script, component, query, api endpoint, class, method, refactor +12 more

  1. +Language / stack & versionInfer from filename/snippet/framework; else pick the modern mainstream default (Python 3.11+, TS+React 18, PostgreSQL) and state the assumption.
  2. +Inputs, outputs & signatureState an explicit typed signature, name it descriptively, and show one example input and its expected output.
  3. +Edge cases & error handlingHandle empty/null/malformed input; choose and state an explicit failure mode (raise vs sentinel).
  4. +Constraints & dependenciesPrefer stdlib / no new deps unless clearly needed; call out any dependency and note complexity if it matters.
  5. +Surrounding code / integration pointWrite self-contained idiomatic code as one pasteable unit; note where it's meant to be called from.
  6. +Output shape (code / explained / tests)Return the complete code first as one block, a brief note on key decisions below, and a couple of usage/test examples.

Left out for this type: role prompting ('senior 10x engineer'); tone/voice instructions ('friendly','professional'); vague 'make it clean, efficient, well-structured'; 'add comments' as a blanket ask; 'use best practices' (name the specific one); over-long company/product backstory; asking for maximal generality on one concrete use

Debugging

Routed here by error, bug, is broken, broken build, broken after, broken test, not working, doesn't work +18 more

  1. _Exact symptom & error textOnly you know this. Paste the full error message and stack trace verbatim, plus observed-vs-expected behavior.
  2. _Reproduction steps & triggerOnly you know this. State the exact steps/input that trigger it and whether it happens every time or intermittently.
  3. _Environment & what changedOnly you know this. Note versions/OS/runtime and the most recent change (deploy, upgrade, config); when it last worked.
  4. _Relevant code & dataOnly you know this. Include the failing function/config and a real sample of the input flowing through it.
  5. _What you already triedOnly you know this. List what you've already checked or attempted and its effect.
  6. +Desired answer shapeRanked list of most-likely causes, each with a concrete way to confirm it and the fix.
  7. +Scope constraintKeep the fix scoped to this bug; don't rewrite unrelated code.

Left out for this type: role/persona preamble ('expert senior engineer'); tone directives (friendly, encouraging); vague 'concise and well-structured'; asking for a full rewrite 'to be safe'; long product/business-goal background; 'explain best practices for X' instead of diagnosing

Creative writing

Routed here by write a story, poem, short story, scene, lyrics, song, flash fiction, monologue +10 more

  1. +Form & lengthName the form precisely and cap it; pick the shortest form that fits (flash ~500 words, poem ~12-20 lines, scene ~1 page).
  2. +POV & tenseClose third-person past for stories; first-person present for intimate/lyric scenes. State it explicitly.
  3. +Tone & moodState 2-3 mood adjectives plus a register, and one 'avoid' to fence off the cliché.
  4. +Concrete anchor / sensory detailSupply one specific setting/object/image and open on it rather than on exposition.
  5. +Ending / turnEnd on an image or a small unresolved beat, not a summary or moral; for poems land on the strongest concrete line.
  6. +Craft constraintsPlain, precise prose; ban the single most obvious cliché for the subject.

Left out for this type: role ('award-winning novelist'); headings/bullets structure; 'make it engaging/vivid/compelling'; 'use rich vocabulary and literary devices'; SEO/audience-persona/CTA scaffolding; 'be creative and original' (a no-op); stating the moral or theme as a thesis

Marketing & ad copy

Routed here by tagline, slogan, landing page, hero copy, headline, product description, ad copy, google ad +12 more

  1. +Target audienceName the single most likely buyer persona and their awareness stage.
  2. +Core value propositionLead with the single most differentiated outcome, phrased as a customer result not a feature.
  3. +What the product actually is / doesState the category and one concrete mechanism; flag specifics to verify rather than invent.
  4. +Differentiator vs alternativesContrast against the status quo and pick one axis (speed, simplicity, price, trust) to own.
  5. +Tone / brand voiceConfident, plain-spoken, benefit-first; no hype words or clichés ('revolutionary','seamless').
  6. +Format & lengthSpecify the exact deliverable and length; provide 3-5 distinct options since copy is chosen by comparison.
  7. +Call to actionInfer the single next step from the funnel stage ('Start free', 'Book a demo').

Left out for this type: role ('world-class copywriter'); vague 'engaging/compelling/catchy/punchy'; generic 'clear and well-structured' format; default long rationale alongside the copy; SEO keyword stuffing where the channel doesn't need it; over-detailed persona backstory; superlatives/buzzwords as tone

Social posts

Routed here by tweet, thread, x post, twitter, linkedin post, linkedin, social post, social media +9 more

  1. +Platform & formatIf unspecified, default to a single tweet/X post under 280 chars (tightest, easiest to expand from).
  2. +Target audienceInfer from the topic's natural community and state the assumption for correction.
  3. +POV / angle / key messageForce one specific, slightly opinionated takeaway rather than a balanced overview.
  4. +Hook / opening lineOpen with a concrete claim or surprising number in the first 7 words; no preamble.
  5. +Objective / desired actionEngagement + credibility: lead with an insight worth reacting to, no hard CTA.
  6. +Voice & toneConversational, first person, confident not hypey; no corporate voice, no emoji spam.
  7. +Hard constraintsRespect the platform char limit; minimal/no hashtags; no emoji unless on-brand; one idea per post.

Left out for this type: role ('viral growth expert'); prose 'well-structured and engaging' format; SEO / keyword-density asks; requesting long background sections; piles of hashtags/emoji as a 'reach' tactic; 'make it go viral' as an instruction; over-specifying reading level or citations

Image-generation prompts

Routed here by image, picture, photo, art, draw, illustration, render, midjourney +13 more

  1. _Subject + specificityOnly you know this. Describe the subject itself — appearance, age or material, what it is doing, what expression it wears. Two or three concrete details.
  2. +Style / mediumInfer from intent: photorealistic for real scenes, clean flat vector for logos, cinematic painting for characters.
  3. +Composition / shot / framingMedium shot, eye-level, subject slightly off-center; close-up for portraits, wide for environments.
  4. +Lighting + moodSoft golden-hour for people/scenes; clean even studio lighting for product/logo work.
  5. +Setting / backgroundAdd a simple fitting environment; plain seamless background for logos/products.
  6. +Color paletteHarmonious, slightly muted palette unless the subject implies specific colors; echo brand colors for brand work.
  7. +Aspect ratioMatch to use: 16:9 scenes/wallpaper, 9:16 phone, 1:1 avatars/logos; default 16:9.
  8. +Detail / render cues'Highly detailed, sharp focus' — a couple of quality cues, not a keyword pile-up.
  9. +Negatives / exclusions'No text, no watermark, no extra limbs, no distortion' as a baseline for figurative work.

Left out for this type: role prompting (persona leaks into the scene as a token); prose format instructions ('respond clearly'); politeness/meta words parsed as content tokens; narrative explanation of why you want it; vague superlatives alone ('make it amazing'); keyword-stuffing dozens of style/quality tags; writing-task tone words ('friendly, conversational')

ASCII & text art

Routed here by ascii art, ascii, text art, banner, figlet, monospace, text logo, box drawing +6 more

  1. +Shape & mediumBlock-letter banner for word/title requests; labeled box-and-arrow diagram for structural/flow requests.
  2. +Exact literal text / labelsUse the exact word/phrase quoted; if none, use a clear placeholder.
  3. +Character set / palettePure 7-bit ASCII unless the target clearly supports Unicode; for clean boxes use box-drawing chars and note the tradeoff.
  4. +Dimensions / width80 columns max; standard figlet height (~6 rows) for banners.
  5. +Render target / where it livesREADME/terminal-safe block wrapped in a fenced code block so monospacing is preserved.
  6. +Raw copy-pasteable outputReturn only the art inside a single fenced code block, nothing else.

Left out for this type: role ('expert ASCII artist'); prose tone/voice descriptors; 'clear and well-structured'; audience persona ('for busy executives'); prose explanation of design choices inside the block; SEO/word-length/reading-level asks; color adjectives when the medium is plain text

Data & analysis

Routed here by sql, query, database, table, dataset, csv, spreadsheet, excel +18 more

  1. +Schema / data shapeAsk for column names + types (or a CREATE TABLE / first rows) and state what one row represents before writing anything.
  2. +Precise analytical questionRestate the vague ask as one measurable question: '[metric] by [dimension] over [window], vs [baseline]'.
  3. +Dialect / tool & engineDefault to standard ANSI SQL (or pandas for files) and state it; ask the warehouse if date/window functions are involved.
  4. +Filters, window & grainState explicit filters (exclude nulls/test/internal), a concrete window, and the rollup grain; surface these as assumptions.
  5. +Edge-case & null handlingHandle NULLs explicitly, guard divide-by-zero, de-dupe on the key; note how each was handled.
  6. +Output shapeProvide the query/formula, a sample of the expected result columns, and a one-line read of what the numbers mean.

Left out for this type: role ('SQL wizard'); 'professional and friendly' tone; vague 'concise and well-structured'; 'make it efficient' with no schema/size/index info; 'be accurate'/'double-check' padding; over-specifying chart styling (a separate viz task)

Research & summarize

Routed here by summarize, summarise, tldr, tl;dr, key points, key takeaways, gist, recap +12 more

  1. _Source materialOnly you know this. Require the pasted/attached text or link; if only a topic was named, flag that no source was supplied rather than inventing content.
  2. +Output format & length5-7 tight bullets, most important first, plus a one-sentence takeaway on top; ~150 words unless the source is long.
  3. +Focus / angleFor a general reader: capture the main thesis, key evidence, and conclusions; ask the angle if stakes are high.
  4. +Audience & purposeAssume an intelligent non-specialist who hasn't seen the source; define jargon briefly, keep it self-contained.
  5. +Fidelity constraintsStick strictly to the source, add nothing, preserve figures/entities exactly, and mark anything ambiguous rather than guessing.
  6. +Comparison dimensionsFor compare tasks, pick 4-6 decision-relevant axes and render a table with a 'which to choose when' verdict.

Left out for this type: role ('expert summarizer'); vague 'concise/comprehensive' adjectives; engaging-tone direction that invites embellishment; asking for 'interesting insights' (adds claims); few-shot examples (a format spec is cheaper); 'read carefully, think step by step' ceremony

Explain & teach

Routed here by explain, eli5, explain like i'm 5, what is, how does, help me understand, teach me, tutorial +9 more

  1. +Audience / prior knowledgeA curious adult with general education but zero domain jargon.
  2. +Depth & scope boundaryEnough for a solid working mental model in a few minutes; skip edge cases and history unless essential.
  3. +Analogy / concrete anchorGround the concept in one everyday analogy, then connect each part of it back to the real thing.
  4. +Format / shapeLead with a one-sentence answer, then a short intuitive build-up; a numbered list for anything sequential.
  5. +Purpose / why they're askingAssume they want durable understanding they could re-explain, not a definition to memorize.
  6. +ConstraintsPlain language, define jargon on first use, and flag where you're deliberately simplifying.

Left out for this type: role ('expert professor'); 'professional and formal' tone (raises reading level); vague 'concise and well-structured'; 'be accurate, don't make things up' filler; rigid word counts (specify depth instead); 'cite sources' for a stable concept

Planning & strategy

Routed here by plan, roadmap, strategy, prioritize, decide, decision, should i, should we +14 more

  1. +Objective & success criteriaState the primary outcome and how you'll know it worked; if unstated, make it explicit and label it an assumption.
  2. +Starting point / contextSummarize what exists, what's decided, and known blockers in 2-3 lines; don't assume greenfield.
  3. +Constraints (time, budget, people, tech)Assume a solo/small team, a named horizon, no new budget; flag these as assumptions the plan hinges on.
  4. +Options + decision criteriaFor decisions, enumerate 2-4 options and score them on cost/time-to-value/risk/reversibility, then recommend one.
  5. +Output shapePhased milestones with a one-line goal and a checkable exit criterion each; a ranked effort/impact table for pure prioritization.
  6. +Risks & dependenciesAdd a short section on the 2-3 things most likely to break the plan, with mitigations and pre-start decisions.

Left out for this type: role ('expert strategist/consultant'); tone directives (formal, engaging); vague 'concise and well-structured'; 'be comprehensive' (bloats to 20 items); prose style anchors; over-specifying every task before the big rocks settle

Difficult conversations

Routed here by negotiate, raise, salary, feedback, give feedback, hard conversation, difficult conversation, confront +14 more

  1. +Relationship & power dynamicAssume a roughly equal peer with ongoing history you want to preserve; flag that it shifts if it's your manager or report.
  2. +Situation & specificsUse 1-2 concrete recent examples with bracketed placeholders the user swaps in for real specifics.
  3. +Desired outcomeOptimize for a specific behavioral or decision change going forward while keeping the relationship intact.
  4. +Tone & registerWarm but direct, plain-spoken, non-defensive; own your part, no groveling and no blaming.
  5. +What to produceA short opening line plus 3-4 flexible talking points to adapt live, not a rigid word-for-word monologue.
  6. +Likely pushback + responsesInclude the 1-2 most likely pushbacks with a calm one-line response each, plus a fallback if it stalls.
  7. +Channel & timingAssume a live 1:1; note that a written version would need to be self-contained and more carefully worded.

Left out for this type: role ('world-class negotiation expert') — biases toward manipulative scripts; vague 'concise and well-structured'; length caps on the guidance (throttle the message, not the advice); 'be professional and respectful' boilerplate; demanding exhaustive background before helping; forcing canned frameworks (SBI/NVC/sandwich) by default

Career docs

Routed here by resume, résumé, cv, cover letter, linkedin, linkedin about, bio, professional bio +10 more

  1. +Target role / job descriptionAsk for the exact title and a pasted JD; else write to the most senior implied title and flag it should be retargeted per application.
  2. +Source facts (experience, wins, metrics)Leave bracketed placeholders like [metric], [team size] rather than inventing; prompt for the top 3 quantifiable wins.
  3. +Artifact + length/formatInfer from the ask: LinkedIn About ~3-4 first-person paras; resume = bullets no pronouns; cover letter ~250-300 words; bio 100-150 words third person.
  4. +Seniority level & industryInfer level from years/scope in the supplied experience; mirror the JD's vocabulary.
  5. +Point of viewResume = no pronouns/past-tense verbs; cover letter & LinkedIn = first person; bio = third person.
  6. +TonePlain, quietly confident, no buzzwords.
  7. +Anti-fabrication constraintNever invent metrics, employers, or dates; missing facts become [placeholders], not fiction.

Left out for this type: role ('world-class career coach'); 'professional and engaging' tone default; 'make it ATS-friendly' as a vibe without the JD; 'well-structured and easy to read' filler; inventing metrics/employers to impress; clichés ('passionate','results-driven','dynamic'); sample-bullet anchors that flatten specificity

Specs & requirements

Routed here by prd, spec, product requirements, requirements doc, ticket, jira ticket, user story, acceptance criteria +8 more

  1. +Deliverable + featureProduce a lightweight PRD: problem, goals, user stories, functional requirements, acceptance criteria, non-goals, open questions.
  2. +Problem & user / why nowState the assumed user, the problem it solves, and the business reason; flag these as assumptions to confirm.
  3. +In-scope vs non-goalsDefine a minimal v1 and an explicit 'Non-goals / later' section for anything adjacent.
  4. +Acceptance criteria / definition of doneWrite testable, checkbox / Given-When-Then criteria covering happy path, empty/error states, and edge cases.
  5. +Constraints (tech, timeline, compliance)Assume the current stack; note timeline and compliance as open questions; never fabricate SLAs or metrics.
  6. +Document shape / sectionsStandard PRD sections with headers, skimmable with a short summary up top; match artifact weight to the ask.
  7. +Prioritization / phasingTag each requirement P0/P1/P2 or split into v1 vs later phases.

Left out for this type: role ('expert product manager'); 'professional and engaging' tone; padding v1 with speculative features; inventing SLAs/metrics as fact; marketing/benefit-selling in the spec body; over-formatting a 2-line ticket into a full PRD

Brainstorming

Routed here by brainstorm, ideas, ideate, name ideas, names for, come up with, generate, list of ideas +14 more

  1. +Context (what it's for)State one line: what the thing is, who it's for, and one distinctive trait.
  2. +QuantityGenerate 15-20 options (volume then selection).
  3. +Diversity / spread of directionsSpread across 3-4 distinct labeled directions (literal/descriptive, evocative, coined/playful, unexpected).
  4. +Constraints / must-avoidsState hard filters (length, tone, clichés to avoid, availability); else avoid the most overused clichés in the space.
  5. +Output shapeOne idea per line, grouped by direction, each with a ≤10-word note on the angle or feel.
  6. +Anchor / example of tasteAsk for one liked/disliked example; else offer a couple of contrasting reference points and name which you're aiming for.

Left out for this type: role ('world-class creative director'); 'be creative / think outside the box'; prose paragraphs instead of a scannable list; over-constraining before divergence (8 filters up front); asking for the single 'best' idea in one shot; deep paragraph rationale per idea

Rewrite & tone-shift

Routed here by rewrite, reword, rephrase, make it formal, make it casual, make it professional, make it sound, tone +17 more

  1. _The actual text to rewriteOnly you know this. Ask for the exact text to rewrite; do not invent or paraphrase from a description.
  2. +Target tone / registerClear and professional but human — plain, direct, warm without being cutesy; no corporate jargon.
  3. +What to preservePreserve all facts, names, numbers, dates, and the core ask/decision exactly; change wording and tone, never meaning.
  4. +Audience & relationshipA professional peer / light-hierarchy recipient you have a normal working relationship with.
  5. +Length / shapeRoughly the same length unless asked to shorten; if shortening, the shortest version that keeps every key point.
  6. +Voice / signature to keepKeep it sounding like a real person — natural contractions, no buzzwords; don't over-formalize into stiffness.

Left out for this type: role ('expert copywriter/editor'); generic 'clear and well-structured'; unrequested multiple creative variations; adding new content/claims ('make it pop'); annotating what you changed unless asked; reading-level / SEO targets; defaulting to 'more formal = better'

Roleplay & persona

Routed here by act as, pretend to be, roleplay, role-play, simulate, you are a, be my, play the role of +11 more

  1. +Persona / who you areA specific, credentialed version of the role (e.g. 'a senior engineering manager who has run 200+ interviews'), not a generic 'expert'.
  2. +Scenario / situationAnchor to the user's real context if known; else pick one concrete common scenario and state it up front for correction.
  3. +Turn-taking rulesAsk one thing at a time, wait for my reply, stay in character, and don't answer on my behalf or monologue.
  4. +Difficulty / realismRealistic and moderately challenging — press on weak answers with follow-ups but stay fair; intermediate level.
  5. +Feedback / evaluationStay in character during; at the end drop character and give scored feedback — strengths, weaknesses, 2-3 improvements, better sample answers.
  6. +Session controls / pause word'pause' steps out to coach or clarify; 'resume' returns where we left off.

Left out for this type: prose 'clear and well-structured' format; 'you are an expert assistant' (competes with the persona); global length caps (fight both short turns and long coaching); 'be comprehensive' (roleplay reveals gradually); citations/sources (break immersion); report-style headers/bullets imposed on dialogue

Speeches & spoken remarks

Routed here by toast, wedding toast, best man speech, maid of honor, eulogy, vows, wedding vows, give a speech +9 more

  1. _Occasion & audienceOnly you know this. State the occasion, who is in the room, and your relationship to the honoree — it sets what is appropriate.
  2. _One concrete storyOnly you know this. Anchor on a single specific, true moment rather than a list of the person’s good qualities.
  3. +Humor registerWarm and affectionate; land one or two light laughs, nothing that would embarrass the honoree.
  4. +Length / time on your feet2 to 3 minutes spoken (~300-400 words); short sentences that read aloud in one breath.
  5. +The close / raise a glassEnd on a direct line to the honoree and a clear toast ("to —").

Left out for this type: dense written-prose sentences that don't say aloud well; a list of generic virtues instead of one real story; inside jokes only part of the room gets; essay or corporate structure; persona prompting ('as a professional speechwriter'); opening with 'the dictionary defines'

Pattern & text-extraction

Routed here by regex, regular expression, grep, ripgrep, glob, jq, sed, awk +9 more

  1. +Tool / dialectName the tool and flavor (JS/PCRE regex, ripgrep, jq, sed); default to PCRE/JS regex and say so.
  2. _Sample input + what matchesOnly you know this. Give 2 to 3 real example lines and mark exactly what should and should NOT match.
  3. +Capture vs matchSay whether you need the whole match or specific capture groups, and in what form.
  4. +Edge casesNote tricky cases: unicode, escaping, multiline, greedy vs lazy, empty input.
  5. +Output formReturn the pattern first, then one line explaining each part; no surrounding program unless asked.

Left out for this type: code-writing moves (function signature, integration point, error handling); role prompting ('senior engineer'); tone / voice; wrapping a one-line pattern in a full script; prose 'clean and well-structured'

Fixed-form poetry

Routed here by haiku, sonnet, limerick, villanelle, tanka, sestina, acrostic, iambic +4 more

  1. +Form & its rulesHonor the named form exactly: haiku 5-7-5 across 3 lines; sonnet 14 lines with a rhyme scheme and meter; limerick AABBA anapestic; etc.
  2. +Subject & concrete imageAnchor on one specific image or moment, not an abstract theme.
  3. +Meter & rhyme fidelityKeep meter and rhyme strict for the form; don't pad lines with filler to hit the count.
  4. +The turn / landingPlace the volta for a sonnet; land the final line as the pivot for a haiku; put the punch in line 5 of a limerick.
  5. +DictionPlain, precise words over poetic filler; earn any archaism.

Left out for this type: the generic 'poem, ~12-20 lines' default; free-verse structure when a strict form was named; padding lines to hit a syllable or rhyme count; role prompting ('as a poet laureate'); explaining the poem after it

Why this is published

A reasonable question: doesn’t giving away the framework give away the tool? No. The framework is where the work starts, not what makes it good. What makes it good is knowing which words real people actually use for each of these — and that comes from the scrubbed map of asks, which is built by use and can’t be copied off a page.

Also, it is more useful to you here than hidden. If all you take from this page is that a cold email needs one ask and not three, that was worth publishing.

Try it on something