Claude prompts for briefs, outlines, drafts and repurpose packs
Copy-ready prompts for the content production path: brief, outline, draft, repurpose, editorial check and batch plan. Paste your notes. Claude must not invent sources, quotes or results you did not supply.
In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Drafted with AI writing assistance and checked against our editorial policy.
Creative Briefs
5 promptsIntake brief from a messy request
1/30✨ What it does
Turns a messy request into a one-page intake brief with labelled gaps. Empty blocks stay tagged until you fill them from a real stakeholder note.
Turn this messy request into a one-page intake brief. Request: [PASTE REQUEST]. Output labelled blocks: Job to do, Audience I can name, Channel, Due date, Assets I already have, Open questions. Quote short phrases from the request. Add no product, campaign or metric I cannot point to in the paste. If a block has no support, write [VERIFY] and leave it empty. Inventing a due date or a channel would send production the wrong way.
Pro tip: Paste the original request, including the messy parts.
Audience brief from your notes
2/30✨ What it does
Writes a short audience brief from notes you already took, including reading context. Missing titles or contexts stay tagged instead of becoming a made-up persona.
Build an audience brief of 180 to 220 words from notes I already took. Notes: [PASTE AUDIENCE NOTES]. Include who they are, what they are trying to finish, what they already tried, and the reading context (phone, desk, commute). Stay inside the notes. If I did not name a job title or a reading context, write [VERIFY] on that line. Do not invent a persona name, a quote or a pain they never stated.
Pro tip: Write notes from one real reader, not a blend of five.
Angle brief with a proof rule
3/30✨ What it does
Builds an angle brief that pairs a promise with proof you can show, plus a will-not-claim line. Thin proof is tagged so you can shrink the promise before drafting.
Write an angle brief for [WORKING TITLE] aimed at [NAMED AUDIENCE]. Promise I will stand behind: [PROMISE]. Proof I can actually show: [PROOF I CAN SHOW]. Constraint I must keep: [CONSTRAINT]. Return four labelled lines: Promise, Proof, Constraint, and What this piece will not claim. If proof is thinner than the promise, write [VERIFY] beside Proof. Never invent a case study, a test result or a named source to prop up the angle.
Pro tip: If proof is a single anecdote, say so. Do not dress it as a study.
Source of truth pack
4/30✨ What it does
Turns canon text, forbidden claims and preferred terms into must-echo, must-avoid and open-fact lists. Unused terms are tagged instead of being treated as published facts.
Make a source of truth pack I can hand a writer. Canon text I trust: [PASTE CANON TEXT]. Claims I forbid: [FORBIDDEN CLAIMS]. Terms I use on purpose: [PREFERRED TERMS]. Output three lists: Must echo, Must avoid, and Open facts. Quote from the canon. If a preferred term never appears in the canon, write [VERIFY] beside it. Do not invent a statistic, a customer quote or a product name to fill Must echo.
Pro tip: Paste the page or memo you actually ship from, not a brainstorm.
Brief gap score before kickoff
5/30✨ What it does
Scores a pasted brief on job, audience, proof, channel and due date, then gives a go or wait line. Missing items are tagged and scored as zero, not filled in.
Score this brief out of 10 on job, audience, proof, channel and due date. Brief: [PASTE BRIEF]. Output a score line, three strengths, three gaps and a go or wait verdict in one sentence. If a scored item is missing from the brief, write [VERIFY] next to that item and cap its score at 0. Do not invent a stakeholder, a launch date or a distribution plan to push a go.
Pro tip: Run this before you outline. A wait is cheaper than a rewrite.
Outlines and Structure
5 promptsArticle outline with word budgets
6/30✨ What it does
Turns a brief into a section table with headings, jobs and word budgets that add up. Sections without a source stay tagged so you collect one before drafting.
Build an outline for [WORKING TITLE] from this brief: [PASTE BRIEF]. Target words: [TARGET WORDS]. Return a table with columns Section, Heading, Job of the section, Word budget and Source I will use. Headings in sentence case. Word budgets must add up to [TARGET WORDS]. Stay inside the brief. If a section has no source in the brief, write [VERIFY] under Source. Never invent a heading that promises a result the brief does not support.
Pro tip: Set the target from a piece you already published, not a guess.
Series outline from one theme
7/30✨ What it does
Lays out a numbered series from one theme note, with each piece answering a question and depending on an earlier one only when needed. Unsupported pieces are tagged, not padded.
Plan a series of [NUMBER OF PIECES] pieces from this theme note: [PASTE THEME NOTE]. Audience: [NAMED AUDIENCE]. Output a table with columns Piece, Working title, Question it answers, Depends on piece, and Status of facts. Piece 1 must stand alone. If the theme note cannot support a piece, write [VERIFY] in Status of facts and do not pad with a filler episode. Inventing a later chapter I did not brief would break the series.
Pro tip: Keep piece 1 useful on its own so a new reader can start there.
Reorder sections for a tired reader
8/30✨ What it does
Reorders your existing sections for a tired reader in a named context, with a short note on what moved. It will not invent a new section or a call to action you did not brief.
Reorder these sections so a tired [NAMED AUDIENCE] can act before they bounce. Current outline: [PASTE OUTLINE]. Reading context: [READING CONTEXT]. Return the new order as a numbered list, then three bullets on what moved and why. Do not add sections. If the outline lacks a next step the reader can take, write [VERIFY] at the end rather than inventing a call to action.
Pro tip: Name the context, such as phone on a commute. Desk readers tolerate more setup.
Heading pass for scan readers
9/30✨ What it does
Marks each heading Keep, Tighten or Replace against the brief promise, with a short rewrite when needed. Headings that claim ungiven proof are tagged and replaced with safer lines.
Critique these headings for a scan reader. Headings: [PASTE HEADINGS]. Brief promise: [PROMISE]. For each heading give Keep, Tighten or Replace, a 12 word reason, and if Tighten or Replace, one rewrite of 8 words or fewer. Stay inside the promise. If a heading claims a proof I did not give, write [VERIFY] and Replace it with a safer heading. Do not invent a number, a tool or a how-to the outline never covered.
Pro tip: Read only the headings first. If the story is unclear, the body will not save it.
Outline versus brief delta
10/30✨ What it does
Compares outline coverage to the brief and marks add, cut or keep on each item. Extra topics you did not allow are cut, and unclear coverage is tagged.
Compare this outline to the brief and list the deltas. Outline: [PASTE OUTLINE]. Brief: [PASTE BRIEF]. Output a table with columns Brief item, Outline coverage, Add cut or keep, and Note under 20 words. If the outline adds a topic the brief never named, mark Cut unless I wrote it in [ALLOWED EXTRAS]. If coverage is unclear, write [VERIFY]. Never invent a brief item to make the outline look complete.
Pro tip: Paste the signed brief, not the hallway version of the request.
First Drafts
5 promptsFirst draft from a locked outline
11/30✨ What it does
Writes a first draft under locked outline headings using only the facts you pasted. Thin sections become tagged stubs instead of padded copy.
Write a first draft from this locked outline and these facts only. Outline: [PASTE OUTLINE]. Facts: [PASTE FACTS]. Voice notes: [VOICE NOTES]. Target words: [TARGET WORDS]. Use the outline headings as H2s. Do not add a new H2. If a section lacks facts, write a one-line stub tagged [VERIFY] and move on. Do not invent a quote, a step, a screenshot or a result to finish a section.
Pro tip: Lock the outline first. Drafting and outlining in one pass hides gaps.
Lede rewrite that stays true
12/30✨ What it does
Offers three 60 to 90 word ledes from one opening fact and the brief promise, then picks one. No draft proceeds if the opening fact is blank.
Rewrite only the lede of this draft, about 60 to 90 words, for [NAMED AUDIENCE]. Draft: [PASTE OPENING]. Fact I may open on: [OPENING FACT]. Promise from the brief: [PROMISE]. Give three ledes, then pick one and say why in 30 words. Each lede must use [OPENING FACT]. If that fact is missing, write [VERIFY] and refuse to draft. Inventing a hook statistic or a scene I did not describe is off limits.
Pro tip: Open on a fact the reader can check, not a mood.
Expand a thin middle section
13/30✨ What it does
Expands one thin section toward a word target using only supporting notes. If the notes run out, it stops and tags what is missing instead of padding.
Expand this thin section to about [TARGET WORDS] words using only supporting notes. Section: [PASTE SECTION]. Notes: [PASTE SUPPORTING NOTES]. Keep the heading. Add examples only if they appear in the notes. If the notes cannot support the target, stop at a honest length, list the missing bits as [VERIFY], and do not invent a walkthrough, a quote or a figure.
Pro tip: Paste interview notes or docs, not a wish list of examples.
Cut a draft to a hard length
14/30✨ What it does
Cuts a draft to a hard word cap while protecting facts you marked must-keep. Missing must-keep facts are tagged, and no new claims appear in the shorter draft.
Cut this draft to [MAX WORDS] words without dropping a fact from [MUST KEEP FACTS]. Draft: [PASTE DRAFT]. Return the cut draft, then a list of cuts grouped as Trimmed line, Moved to later, or Dropped. If a must-keep fact is not in the draft, write [VERIFY] and do not invent a replacement sentence. Do not add new claims while cutting.
Pro tip: Mark must-keep facts before you cut, or the trim will eat the point.
Draft fidelity check against the brief
15/30✨ What it does
Maps each brief promise to quoted draft evidence and marks kept or drifted. Promises with no evidence stay tagged, and no rescue paragraph is invented.
Check this draft against the brief and report fidelity. Draft: [PASTE DRAFT]. Brief: [PASTE BRIEF]. Output a table with columns Brief promise, Draft evidence, Kept or drifted, and Fix in 15 words. Quote the draft. If a promise has no evidence, write [VERIFY] under Draft evidence. Do not invent a paragraph that would bring the draft back on brief.
Pro tip: Run this before design. Drift is cheaper to fix in text.
These prompts give you the what. Tutorials give you the why.
Learn when to use extended thinking, how to build Claude Projects, and workflows that compound. The library keeps growing.
Repurpose Packs
5 promptsArticle to email pack
16/30✨ What it does
Turns one article into a short email and a 6 to 8 word subject, using only article facts. Unsupported reader actions are tagged rather than dressed up with a fake bonus.
Turn this article into one email of 140 to 180 words plus a subject line of 6 to 8 words. Article: [PASTE ARTICLE]. Reader action I want: [READER ACTION]. Audience: [NAMED AUDIENCE]. Use only facts in the article. If the action is not supported by the article, write [VERIFY] on the last line and do not invent a bonus, a deadline or a download.
Pro tip: Pick one action. Two actions in one email split the click.
Five posts from one piece
17/30✨ What it does
Splits one piece into five numbered posts with a short hook, body and next step. Posts that need a missing result, name or date are tagged and skipped.
Split this piece into five posts. Piece: [PASTE PIECE]. Channel: [CHANNEL]. For each post give a hook under 12 words, a body of 50 to 80 words, and one next step. Stay inside the piece. If a result, name or date is missing, write [VERIFY] in that post and skip it rather than inventing a proof point. Number the posts 1 to 5.
Pro tip: Assign each post a different section so they do not clone each other.
Thread from a draft section
18/30✨ What it does
Turns one section into a numbered thread with a clear first post and a last-post action only when the section supports it. Short sections stop early instead of being padded.
Turn this section into a thread of [NUMBER OF POSTS] posts for [CHANNEL]. Section: [PASTE SECTION]. Each post 40 to 70 words. Post 1 states the point. Last post states [READER ACTION] only if the section supports it. If the section cannot fill the count, stop early and tag remaining slots [VERIFY]. Do not invent a story beat, a screenshot or a quote to reach the count.
Pro tip: Count posts from the section's real beats, then set the number.
Talk track from a published piece
19/30✨ What it does
Writes a 90 second, four-part talk track from a published piece, sized near 195 words. Missing figures stay tagged, and no live demo or client is added.
Write a 90 second talk track from this published piece for a [EVENT TYPE] audience. Piece: [PASTE PIECE]. Speech runs at about 130 words a minute, so keep the track near 195 words. Use four labelled parts: hook, problem, one teaching point, close. Close with [READER ACTION] only if the piece states it. If a teaching point needs a figure the piece lacks, write [VERIFY] there. Never invent a live demo, a client name or a result.
Pro tip: Read it aloud once. Cut any sentence you cannot say in one breath.
Newsletter blurb from a longer draft
20/30✨ What it does
Compresses a longer draft into a 70 to 90 word newsletter blurb and a short teaser. Missing next sentences are tagged, and no guest or metric is added for freshness.
Write a newsletter blurb of 70 to 90 words from this longer draft, plus a 5 to 7 word teaser line. Draft: [PASTE DRAFT]. Link label I will use: [LINK LABEL]. Audience: [NAMED AUDIENCE]. Stay inside the draft. If the draft has no concrete next sentence for the reader, write [VERIFY] at the end. Do not invent an issue theme, a guest or a metric to make the blurb feel timely.
Pro tip: Use the same link label in the draft and in the email so tracking stays honest.
Editorial Review
5 promptsClaim check against your notes
21/30✨ What it does
Pulls verifiable claims from a draft and marks keep, hedge or cut against notes you trust. Unsupported claims are tagged and cut, not rescued with a fake source.
Extract every claim from this draft that a reader might try to verify. Draft: [PASTE DRAFT]. Notes I trust: [PASTE NOTES]. Output a table with columns Claim quoted, Supported in notes, and Action (keep, hedge, or cut). If a claim is not in the notes, write [VERIFY] and mark cut. Do not invent a source, a study or an internal doc to save a line.
Pro tip: Paste the notes the writer actually used, not a later research pile.
Tone pass for one audience
22/30✨ What it does
Scores a draft on warmth, plain words, specificity and respect, then rewrites the weakest paragraph. Missing voice rules are tagged rather than invented.
Score this draft from 1 to 10 on warmth, plain words, specificity and respect for [NAMED AUDIENCE]. Draft: [PASTE DRAFT]. Voice notes: [VOICE NOTES]. Output the four scores, one rewrite of the weakest paragraph, and a list of words to swap. Do not add facts. If a score depends on a voice rule I did not give, write [VERIFY] beside that score. Inventing a brand voice rule I did not write is not allowed.
Pro tip: Name the audience as a real role, such as a clinic manager, not everyone.
Call to action check
23/30✨ What it does
Quotes each call to action and marks match, drift or missing against the one action you allow. Extra asks are tagged, and no destination is invented.
Audit the calls to action in this draft. Draft: [PASTE DRAFT]. Allowed action: [ALLOWED ACTION]. Destination I control: [DESTINATION]. List each CTA quoted, then mark Match, Drift or Missing. If the draft asks for an action I did not allow, mark Drift and write [VERIFY]. Do not invent a button label, a URL or a lead magnet to complete a missing CTA.
Pro tip: One allowed action per piece. Extra asks hide the click you wanted.
Source list the reader can follow
24/30✨ What it does
Builds a numbered source list from materials you used, tied to the draft. Claims with no matching material get a tagged row, and no publication is invented.
Build a source list from materials I actually used. Materials: [PASTE SOURCE NOTES]. Draft: [PASTE DRAFT]. Output a numbered list with Title or label, What I used it for, and Where the reader can find it if I supplied a location. If a draft claim has no matching material, write [VERIFY] on a new row. Never invent a publication, a URL or an author.
Pro tip: If you cannot say where a reader finds it, it is not a source yet.
Fact gap table before publish
25/30✨ What it does
Lists dates, names, figures, product behaviour and quotes with a confirm-or-gap status. Gaps stay tagged for the owner you named, and no figure is filled in.
Make a fact gap table I can clear before publish. Draft: [PASTE DRAFT]. Owner who can confirm: [OWNER ROLE]. Output rows for dates, names, figures, product behaviour and quotes. Columns Fact, Location in draft, Status (ok or gap), and Who confirms. Status is gap and tagged [VERIFY] when I did not supply a confirmation. Do not invent an owner, a figure or a quote to close a row.
Pro tip: Walk the table with the owner on a call, not in a silent comment thread.
Most people use 10% of Claude. Tutorials unlock the rest.
AI Academy: hands-on tutorials on Claude, ChatGPT, Midjourney, and many other AI tools. New tutorials are added regularly.
Batch and Handoff
5 promptsOne week production board
26/30✨ What it does
Turns your hours and in-flight pieces into a Monday to Friday board with stages and blockers. Missing stages are tagged, and no extra reviewer is added.
Build a Monday to Friday production board for [WEEK DATES]. Hours I have: [HOURS AVAILABLE]. Pieces in flight: [PASTE PIECE LIST]. Output a table with columns Day, Piece, Stage (brief, outline, draft, edit, ship), Hours and Blocker. Hours must not exceed [HOURS AVAILABLE] across the week. If a piece has no stage, write [VERIFY]. Do not invent a ship date, a designer or a review slot I did not list.
Pro tip: Count hours from last week, including the meetings that ate them.
Asset list for one piece
27/30✨ What it does
Lists remaining assets for one piece with owners and dates you already wrote. Implied assets without an owner stay tagged, and no photographer or stock shot is invented.
List the assets this piece still needs. Piece notes: [PASTE PIECE NOTES]. Output a table with columns Asset, Owner I named, Need by, and Status. Use only owners and dates I wrote. If an asset is implied but has no owner, write [VERIFY] under Owner. Never invent a photographer, a tool login or a stock shot to complete the list.
Pro tip: Walk the draft once and list every image, clip and pull quote.
Handoff note to an editor
28/30✨ What it does
Drafts a 130 word editor handoff with brief location, known gaps and the decision you still own. Unowned gaps are tagged, and no legal review is invented.
Write a handoff note of about 130 words to an editor for [WORKING TITLE]. Include where the brief lives in [BRIEF LOCATION], known gaps tagged from [KNOWN GAPS], and the one decision I still own. Tone: direct, no apology padding. If a gap has no owner, write [VERIFY]. Do not invent a style guide rule, a legal review or a publish slot.
Pro tip: Name the one decision you still own so the editor does not guess.
Status table for a stuck pack
29/30✨ What it does
Turns messy status notes into a one-minute scan table with last done, stuck on and next action. Missing dates are tagged, and no fake progress appears.
Turn these status notes into a table a manager can scan in one minute. Notes: [PASTE STATUS NOTES]. Pieces: [PASTE PIECE NAMES]. Columns Piece, Last done, Stuck on, Next action, and Date I already set. Next action must be a verb plus object from the notes. If a date is missing, write [VERIFY]. Do not invent progress, a reviewer sign-off or a ship.
Pro tip: Paste Slack scraps and doc comments together so blockers surface.
Kill list for weak ideas
30/30✨ What it does
Marks each idea Keep or Kill against the brief and remaining hours, with a short reason. Vague ideas are tagged, and no trend is invented to save them.
From this idea list, mark which to kill this month. Ideas: [PASTE IDEA LIST]. Brief we are serving: [PASTE BRIEF]. Hours left: [HOURS LEFT]. Output a table with columns Idea, Keep or Kill, Reason under 20 words. Kill anything that needs proof I did not attach. If an idea is too vague to judge, write [VERIFY]. Inventing a trend or a request from sales to save an idea is not allowed.
Pro tip: Kill is a production tool. A long maybe list is how weeks vanish.
Free tool
Prompt Optimizer
Turn a rough idea into a structured, professional AI prompt.
Frequently Asked Questions
Prompts are the starting line. Tutorials are the finish.
A growing library of hands-on tutorials on ChatGPT, Claude, Midjourney, and many other AI tools. New tutorials are added regularly.