AI Prompts Library

Design my 30-day learning roadmap for {SKILL — e.g. Python, chess, copywriting, video editing,
Spanish}. Starting level: {TRUE BEGINNER / DABBLED / RUSTY}. Time: {MINUTES/DAY}. My goal in 30 days: {SPECIFIC — "build X", "hold a conversation", "edit one video"}.
Build:
1. THE 20%: the minimal subset of this skill that delivers 80% of my goal — and, just as
important, the list of what to deliberately IGNORE for 30 days (advanced topics that feel
productive but stall beginners).
2. WEEKLY ARCS with a shippable Friday project each week (proof of progress beats streaks): Week 1 = fundamentals + embarrassingly small project; Week 2 = core technique + real project; Week 3 = the hard part everyone avoids + project using it; Week 4 = integration + the goal project.
3. DAILY LOOP ({MY MINUTES}): warm-up retrieval of yesterday (20%) → new material (30%) → hands-on practice (50%). Practice means producing, not consuming — specify the exercise type per week.
4. FEEDBACK MECHANISM: how I'll know I'm doing it wrong each week (the self-check, the community to post in, the comparison test).
5. THE WALL: predict the exact day/week I'll want to quit, why it happens with this skill, and the pre-planned response.
6. DAY 30 EXAM: how I'll prove the goal — pass/fail criteria written today.

Transform my raw {LECTURE NOTES / SLIDES / TRANSCRIPT / TEXTBOOK PHOTOS} into study-grade notes.
Course: {SUBJECT + LEVEL}. Exam style if known: {MCQ / ESSAY / PROBLEMS}.
Produce in Cornell-style structure:
1. NOTES COLUMN: the material reorganized by CONCEPT (not by the order the professor rambled) — hierarchical bullets, definitions boxed, every formula with variables named and units, every
process as numbered steps.
2. CUE COLUMN: for each section, 2–3 retrieval questions (the questions the notes answer — I'll cover the notes and quiz myself). Match my exam style: {MCQ-style distinctions / essay-style "compare and evaluate" / problem-type triggers}.
3. SUMMARY (bottom): the whole lecture in 5 sentences a smart friend would understand.
4. FLAGGED GAPS: where my notes are ambiguous, contradictory, or clearly missing a step — listed as questions to ask the professor or check the textbook. Do NOT silently fill gaps with your own assumptions; mark anything you added as [AI-ADDED — verify].
5. EXAM RADAR: the 3 items here most likely to appear on the exam, and why (emphasized, repeated, or classically testable).
MY RAW MATERIAL: {PASTE}

Build a concept map of {TOPIC / PASTE CHAPTER}. Understanding lives in the CONNECTIONS —
flashcards test dots, exams test the lines between them.
Produce:
1. THE CORE: the single central concept everything else hangs on, and the one-sentence version of it.
2. THE MAP (text form):
CENTRAL CONCEPT
├── Branch concept — [connecting phrase: "causes / requires / is measured by / contrasts with"]
│ ├── sub-concept — [link phrase]
Every line MUST have a labeled relationship — an unlabeled line is a fact you don't actually
understand.
3. CROSS-LINKS: the 5 non-obvious connections BETWEEN branches (these are where exam questions
live: "how does X relate to Y?"). Explain each in one sentence.
4. THE MISCONCEPTION EDGES: 3 pairs that students wrongly connect or conflate — mark them with the distinction that separates them.
5. SELF-TEST: hand me the map skeleton with 8 blanks (mixed concepts AND link-labels) to fill from memory.
6. THE STORY PASS: walk the entire map as one flowing 150-word narrative — if the map can't be
told as a story, the structure is wrong.

Build my study schedule. Exam: {SUBJECT + DATE}. Material to cover: {LIST TOPICS/CHAPTERS + which feel weak vs. strong}. My honest availability: {HOURS/DAY + which days off}. My track record: {I CRAM / I START STRONG THEN FADE / I'M STEADY}.
Build with real learning science:
1. BACKWARDS PLAN: final week = practice exams + weak-spot repair ONLY (no new material);
everything new finishes 7+ days before the exam.
2. SPACED REPETITION built in: every topic reappears at day 1, 3, 7, and 14+ after first study — as RETRIEVAL (self-quiz, blank-page recall), never rereading. Schedule these reviews explicitly.
3. INTERLEAVING: mix 2–3 topics per day rather than one topic per week — harder, sticks better.
4. WEIGHT BY WEAKNESS: weak topics get prime slots (my best hours) and more reps; strong topics get maintenance only.
5. THE HUMAN FACTOR: given my track record, build the failure-resistant version — buffer days, a "minimum viable day" (30 min) for bad days, and one full day off per week (burned out ≠ prepared).
Output: day-by-day table (date | topics | mode: learn/retrieve/practice-test | minutes) + the 3 rules to not break.

Turn my messy week notes into a status report my {MANAGER / CLIENT / TEAM} will actually read.
Report cadence: {WEEKLY / BIWEEKLY}. Their priority right now: {WHAT THEY CARE ABOUT MOST}.
Format:
1. HEADLINE: one sentence — the week's most important truth (progress or problem, whichever is
bigger). Managers often read only this; make it carry the week.
2. SHIPPED/DONE: 3–5 items max, outcome-phrased ("Launched X — early signups: 43") not
activity-phrased ("worked on X"). Group small stuff as one line.
3. IN PROGRESS: what's moving, each with a % or next milestone + date. No vague "ongoing".
4. BLOCKED/NEEDS DECISION: the section that makes reports valuable — each blocker with: what's
stuck, who/what unblocks it, and the specific ask ("Need your yes/no on X by Thursday or the
timeline slips a week"). Never bury an ask in prose.
4. NEXT WEEK: top 3 priorities, so misalignment surfaces NOW instead of next Friday.
Rules: under 200 words total, no adjectives doing the work of numbers, bad news stated plainly in the first half — buried bad news destroys trust when found.
MY NOTES: {PASTE}

Review this contract before I sign it. I'm the {FREELANCER / CLIENT / EMPLOYEE / VENDOR}. Deal
context: {WHAT'S BEING EXCHANGED + ROUGH VALUE}. My biggest worry: {IF ANY}.
Scan for and explain in plain English:
1. RED FLAGS ranked by severity: unlimited liability, IP assignment broader than the work,
non-compete scope/duration, payment terms beyond 30 days, unilateral termination or change rights, auto-renewal traps, indemnification imbalances, "work for hire" surprises. For each: quote the clause → what it means for ME in one sentence → what it could cost in a realistic scenario.
2. WHAT'S MISSING: protections I'd expect for my side that aren't there (kill fee, late-payment interest, revision limits, liability cap).
3. THE ASKS: for the top 3 issues — the specific replacement wording to request, and the
negotiation line to deliver it ("Standard in my agreements is X — can we align on that?").
4. WALK-AWAY TEST: is anything here a dealbreaker at this deal size, or is it all negotiable
friction?
End with the reminder: you're an AI, not my lawyer — for {HIGH-VALUE / UNUSUAL} deals, a real
lawyer reviews the final version.
CONTRACT: {PASTE}

Write my performance self-assessment. My role: {TITLE}. Review covers: {PERIOD}. My raw material:
{PASTE — messy list of what I did, projects, wins, numbers if I have them, and anything that went sideways}.
Rules:
1. IMPACT LANGUAGE: transform every "I did X" into "I did X, which resulted in Y" — where Y is
money, time, risk, quality or team capability. Where I gave no number, mark {ADD METRIC?} and
suggest what to measure.
2. THE HIERARCHY: lead with the 2–3 things that mattered most to the BUSINESS (not the hardest
work — the highest impact; they're often different).
3. HIDDEN WORK: surface the invisible labor — mentoring, unblocking others, preventing fires,
documentation — framed as impact, because unclaimed work doesn't exist at review time.
4. THE GROWTH SECTION: one real area to improve, framed with evidence I'm already addressing it (never a humble-brag "I work too hard", never a real self-torpedo).
5. FORWARD CASE: 3 sentences connecting this year's evidence to next year's scope — the quiet
promotion argument.
Tone: confident, factual, zero arrogance, zero apology. Length: {FORM'S LIMIT or 400 words}.

Create my client onboarding kit. My service: {WHAT I DO}. Typical project: {SCOPE + DURATION +
PRICE RANGE}. Past pain (be honest): {WHAT HAS GONE WRONG BEFORE — scope creep, slow feedback,
ghosting, endless revisions}.
Deliver:
1. ONBOARDING QUESTIONNAIRE: 10–12 questions that actually shape the work — goals with numbers, decision-maker mapping ("who else approves this?"), taste calibration ("show me 2 examples you love and 1 you hate"), constraints, and the trap-detector question for my specific past pain.
2. KICKOFF CALL AGENDA (45 min): introductions kept short → goals confirmed back to them → process walkthrough with THEIR responsibilities explicit (feedback windows, asset delivery) →
communication rules (channel, response times, meeting cadence) → next 2 weeks concrete → the "how we handle changes" conversation NOW, while everyone's happy.
3. THE EXPECTATIONS EMAIL (sent after kickoff): summarizes decisions, deadlines both ways,
revision policy, and the scope boundary — friendly tone, contractual precision. Under 250 words.
4. RED-FLAG PROTOCOL: the 3 early warning signs from my past pain, and the pre-written
gentle-but-firm message for each.

Mine these customer reviews for marketing gold. Source: {MY REVIEWS / COMPETITOR'S REVIEWS from
G2, Amazon, Trustpilot, app store}. My product context: {WHAT I SELL}.
Extract:
1. VERBATIM GOLD: the 10 most vivid customer phrases, quoted exactly — the words they use for the problem ("I was drowning in...") and the result ("finally felt like..."). These become headlines; customers write better copy than copywriters.
2. BUYING TRIGGERS: what event pushed them to buy? Cluster the patterns ("switched after X broke", "new job forced...", "recommended by...").
3. HESITATION PATTERNS: "I almost didn't buy because..." — each one is an objection my page must answer. List objection → the review line that answers it.
4. UNEXPECTED VALUE: what do they praise that the marketing doesn't even mention? (Often the real differentiator.)
5. COMPLAINT CLUSTERS (if competitor reviews): the top 3 repeated frustrations = my positioning openings. Write the comparison-page line for each.
6. READY-TO-USE OUTPUT: 5 headlines, 3 ad opening lines, and 5 FAQ entries — all built from actual customer language, marked with which review inspired each.
REVIEWS: {PASTE — more is better}

Optimize my {APP STORE / GOOGLE PLAY / CHROME WEB STORE / MARKETPLACE} listing. Product: {WHAT IT DOES}. Current listing (if live): {PASTE}. Top 3 competitors: {NAMES}.
Deliver:
1. TITLE + SUBTITLE: within character limits ({30/50} chars), lead with the highest-intent keyword phrase people actually search, brand second. 3 options.
2. KEYWORD STRATEGY: 15 target keywords grouped by intent (problem searches "stop procrastinating" / solution searches "pomodoro timer" / competitor searches). Where each belongs: title, subtitle, keyword field, or description.
3. THE FIRST TWO LINES of the description (visible before "more"): outcome + differentiator, no throat-clearing. 3 options.
4. FULL DESCRIPTION: scannable structure — outcome paragraph → 5 benefit bullets (outcome first, feature second) → social proof line → objection killer (privacy/price/complexity, whichever fits) → CTA.
5. SCREENSHOT STORY: the 5-screenshot sequence as a narrative (hook claim → core action → result → social proof → CTA), with the 4-word caption for each. Screenshot 1 gets 80% of attention — it must work alone.
6. RATINGS PLAY: the in-app moment to ask for a review (after a success, never at open), and the ask copy.

Audit and rewrite my pricing page. My product: {WHAT IT IS}. Tiers and prices: {PASTE}. Target
customer: {WHO}. Current problem: {LOW CONVERSIONS / WRONG TIER CHOSEN / CART ABANDONMENT / JUST LAUNCHED}.
Audit against pricing psychology:
1. ANCHORING: is the order and visual weight guiding people to the RIGHT tier? (Highlight the
middle/target tier; the premium tier's job is often to make it look reasonable.)
2. TIER NAMES: do names describe the CUSTOMER ("Freelancer / Studio / Agency") rather than sizes ("Basic / Pro")? Self-identification sells.
3. FEATURE LISTS: cut features nobody chooses BY; reframe remaining ones as outcomes ("Unlimited projects" → "Never tell a client no").
4. PRICE PRESENTATION: annual/monthly framing, the per-day reframe if honest, and whether the
currency/number formatting adds friction.
5. OBJECTION LAYER: the 3 fears at the buy moment for MY product — where's the guarantee, the
"cancel anytime", the "who is this NOT for"?
6. CTA: button copy per tier (not three "Sign up"s).
Deliver: scored audit (each element ✓/✗ with a one-liner) → fully rewritten page copy → the ONE A/B test to run first and the metric that decides it.

Build my 30-day content calendar. Business: {WHAT I DO + WHO FOR}. Platform priority: {MAIN +
SECONDARY}. Goal this month: {AWARENESS / LEADS / LAUNCH / AUTHORITY}. Time budget: {POSTS PER
WEEK I can realistically sustain}.
Structure:
1. PILLARS: 4 content pillars with the % split — educational (teach one thing), proof (results, testimonials, behind-the-scenes), personal (story, opinion), promotional (max 20%).
2. THE CALENDAR: for each posting day — pillar, format (text / carousel / short video / poll), the specific topic, the HOOK (actual first line, not "hook about X"), core point in one sentence, CTA (rotate: follow / save / comment prompt / link — never the same twice in a row).
3. THE REPURPOSE MAP: how each week's best piece becomes 2 more (long post → 3 short takes; video → text thread).
4. BATCH PLAN: group the month into 4 creation sessions with what to make in each.
5. THE BENCH: 5 evergreen backup topics for days when the planned post feels wrong.

Diagnose the performance of my {CODE / SQL QUERY / ENDPOINT}. Symptoms: {WHAT'S SLOW, HOW SLOW, AT WHAT SCALE — "fine with 100 rows, dies at 100k"}. Environment: {LANGUAGE/DB + anything relevant}.
Method — measure before optimizing:
1. INSTRUMENT FIRST: tell me exactly how to measure where time goes (the profiler command, the
EXPLAIN ANALYZE, the timing wrapper) so we're not guessing.
2. HYPOTHESES: the 3 most likely bottlenecks for this pattern and scale, ranked — with the
tell-tale signature each would show in the measurement.
3. Once identified (or for your top hypothesis now), THE FIX: corrected code/query, with expected order-of-improvement ("this turns O(n²) into O(n log n) — at your 100k rows, roughly 1000× fewer operations").
4. THE HIERARCHY check — in order, because each level dwarfs the next: algorithm choice → I/O and round-trips (N+1 queries, network calls in loops) → memory allocation patterns →
micro-optimizations. Refuse to micro-optimize if a higher level is unfixed.
5. GUARD: a simple benchmark/assertion to keep in place so this regression can't sneak back.
CODE/QUERY: {PASTE}

Decode this error for me. Context: {LANGUAGE/FRAMEWORK + what I was doing when it happened + what
I changed recently, if anything}.
Structure your answer:
1. TRANSLATION: what this error message actually says, in one plain-English sentence. (Error
messages describe symptoms — name the symptom.)
2. WHERE TO LOOK: which line/frame in the trace is MINE vs. library noise — point at the frame
that matters.
3. LIKELY CAUSES, ranked by probability for MY context: for each — why it produces exactly this message, and a 30-second check to confirm or eliminate it (a print, a command, an inspection).
4. THE FIX for the confirmed cause — and if we can't confirm yet, the fix for cause #1 with a note on what result means "move to cause #2".
5. UNDERSTAND IT ONCE: the 2-sentence concept behind this error class, so I recognize it instantly next time.
Rule: never say "this usually means" without committing to a ranked list. Vague error help is no help.
ERROR + STACK TRACE: {PASTE}
RELEVANT CODE (if I have it): {PASTE}

Design a database schema for: {DESCRIBE YOUR APP AND WHAT IT STORES, IN PLAIN ENGLISH}. Database: {PostgreSQL / MySQL / SQLite / "you pick"}. Expected scale: {ROUGH USERS/ROWS or "small, might grow"}.
Process:
1. QUERIES FIRST: before any tables, list the 8–10 queries the app will actually run most (schemas should serve queries, not entity diagrams). Confirm with me if unsure.
2. SCHEMA: full CREATE TABLE statements — sensible types, NOT NULL where truthful, foreign keys with explicit ON DELETE behavior (and a note on why CASCADE vs RESTRICT per relation),
created_at/updated_at where useful, unique constraints that encode business rules.
3. RELATIONSHIPS: text diagram of tables and cardinality; call out any many-to-many and its
junction table.
4. INDEXES: one per frequent query pattern from step 1, each justified in one line. No speculative indexes.
5. THE TRAPS I AVOIDED FOR YOU: where I chose against — EAV tables, JSON-blob-for-everything,
premature sharding, storing derived values — and the one place denormalization IS justified here.
6. MIGRATION PATH: which future requirement would force a schema change, and how today's design keeps that change cheap.

Translate my code from {SOURCE LANGUAGE} to {TARGET LANGUAGE} — as a senior {TARGET} developer
would WRITE it, not as a machine would transliterate it.
Rules:
1. IDIOMATIC FIRST: use {TARGET}'s native patterns (its error handling style, its iteration
idioms, its naming conventions) even when the structure diverges from my original. Word-for-word translations are how you spot a tourist.
2. ECOSYSTEM: replace hand-rolled logic with the standard library or de-facto standard package a native would reach for. Name the package and why it's the community default.
3. SEMANTICS WATCHLIST: after the code, list every place where behavior subtly differs — integer division, string encoding, null vs undefined vs None semantics, mutability, concurrency model,number precision. These bite silently.
4. THE UNTRANSLATABLE: anything with no clean equivalent — explain the trade-off of each
workaround option.
5. TESTS: translate/adapt one test that proves both versions behave identically on the same
inputs, including one edge case from the watchlist.
CODE: {PASTE}

Diagnose my story's structure. It's a {PERSONAL ESSAY / TALK / BLOG NARRATIVE / SHORT STORY / CASE STUDY} and it feels {FLAT / SLOW / POINTLESS / RUSHED — my own diagnosis if I have one}.
Examine against the load-bearing elements:
1. STAKES: what does the protagonist stand to lose? If nothing, that's the flatness. Where can
stakes be raised or revealed earlier?
2. DESIRE vs. OBSTACLE: what do they want, what's in the way? Is the obstacle worthy (external AND internal)?
3. THE TURN: the moment something changes irreversibly. Does mine have one? Is it buried, or does it arrive too easily?
4. SCENE vs. SUMMARY: mark where I'm summarizing ("things got harder") where I should be in-scene (dialogue, sensory detail, real time). The most important beat must be a scene.
5. OPENING/ENDING PAIR: does the ending answer something the opening raised? The best endings are surprising-yet-inevitable.
Deliver: the ONE structural problem hurting it most → the specific fix → rewrite my weakest
section to demonstrate → then a beat-sheet of my story as it SHOULD flow.
MY DRAFT: {PASTE}

Write a layered explainer of {COMPLEX TOPIC} in the "5 levels" format — each level true, none
contradicting the others, each adding resolution.
LEVEL 1 — CHILD (5yo): one everyday analogy, 3 sentences, zero jargon.
LEVEL 2 — TEENAGER: the actual mechanism in plain words, 5–6 sentences, one "huh, cool" fact.
LEVEL 3 — COLLEGE STUDENT: proper terminology (each term glossed at first use), the why behind the mechanism, the main exception or edge case.
LEVEL 4 — GRAD STUDENT: the current debates, what the field got wrong before, quantities and
trade-offs.
LEVEL 5 — EXPERT: the open problems, the frontier, what would change the field if solved.
Rules: each level stands alone; the analogy from level 1 should echo at least once in later levels (continuity makes it memorable); at each level include one thing the PREVIOUS level oversimplified and gently correct it ("earlier I said X — that's almost true, but...").
End with: the one-sentence version someone repeats at dinner, and 3 sources to go deeper.

Write a {WEDDING TOAST / FAREWELL SPEECH / BIRTHDAY SPEECH / RETIREMENT TRIBUTE} for {WHO — my
relationship to them}, {LENGTH: 2–3} minutes, tone: {WARM+FUNNY / HEARTFELT / LIGHT}.
First, interview me — ask one at a time:
1. One specific story that captures who this person is (the smaller and more specific, the better — "the IKEA shelf incident" beats "they're always helpful")
2. One genuine quality everyone in the room would nod at
3. One funny-but-kind detail or running joke (nothing embarrassing — I have to live with these
people)
4. The relationship moment that matters to ME
Then build: cold open that isn't "for those who don't know me" → the story, told with dialogue and detail (this is 60% of the speech) → the quality it reveals → the turn to sincerity (one honest sentence that risks a lump in the throat) → the raise-your-glass line.
Delivery notes: [PAUSE] marks, the line to slow down on, and the reminder: 2 minutes performed
well beats 5 minutes performed nervously.

Write a 5-email welcome sequence for new subscribers to {MY NEWSLETTER/LEAD MAGNET}. My business:
{WHAT I DO}. Eventual paid offer: {PRODUCT/SERVICE}. Voice: {FRIENDLY EXPERT / WITTY / CALM
PROFESSIONAL}.
The sequence (with send timing):
EMAIL 1 (instantly): deliver what they signed up for + one unexpected bonus tip + set expectations ("every Tuesday, 3-minute read") + one question that invites a reply (replies train inbox providers to trust me).
EMAIL 2 (day 2): my story — why I do this, the struggle that made me credible, in 150 words.
Humans subscribe to humans.
EMAIL 3 (day 4): pure value — my single best actionable insight, complete in the email (no "click to read more" tease).
EMAIL 4 (day 7): social proof + the problem my paid offer solves, framed as a story about someone like them. Soft mention, no hard pitch.
EMAIL 5 (day 10): the direct-but-graceful pitch — what I offer, who it's for, who it's NOT for
(honesty converts), one CTA.
Per email: subject (max 45 chars, curiosity or benefit) + preview text + body under 200 words.

Write a kinetic typography promo video script for {ANNOUNCEMENT/PRODUCT/EVENT}, {15/30}s, for
{PLATFORM}.
Rules of the format — the text is the star:
1. THE SCRIPT: 25–45 words TOTAL, broken into 8–14 screen "beats" of 1–5 words each. Each beat = one screen. Write for rhythm: vary beat length (long-short-short-long), build tension, land the payoff word ALONE on screen.
2. EMPHASIS MAP: mark each beat — [BIG] for power words (full-screen, heavy weight), [fast] for connective phrases (quick flashes), [HOLD] for the beats that need a full second, [COLOR] for the 2–3 words in accent color.
3. MOTION NOTES: per beat — slam in / typewriter / slide up / scale punch. Never the same
transition twice in a row.
4. SOUND: where the beat drops, where a whoosh or click lands, where silence hits before the final word.
5. FINAL FRAME: logo + CTA + the one line that stays on screen 2 full seconds.
Give me 2 versions: hype energy and premium-calm energy.

Plan a cinematic walkthrough video of {PROPERTY / PRODUCT / SPACE / VENUE}, target length
{30–90}s, for {LISTING / SOCIAL / WEBSITE HERO}.
Deliver:
1. SHOT SEQUENCE (8–12 shots): the reveal order matters — start with the strongest
exterior/overview beauty shot, then a natural discovery path, save the "wow" feature for 60%
through, end wide. For each shot: duration, camera movement (slow push-in / lateral glide / reveal around corner / rise), and what feeling it sells.
2. AI-GENERATION VERSION: if I'm using AI video from stills — one paste-ready prompt per shot:
"Cinematic {interior/exterior} shot, [scene description], camera slowly {MOVEMENT}, {LIGHTING}, photorealistic, smooth stabilized motion, no morphing, no people, 16:9".
3. PACING & MUSIC: where cuts land (every 3–5s early, longer holds on hero shots), music mood, and the one moment to sync a cut to the beat.
4. TEXT OVERLAYS: max 4 (location/price/specs/CTA), timed.
5. PHONE FALLBACK: how to shoot each move with a phone gimbal if AI shots fail.

Script one course video lesson. Course: {COURSE TOPIC}. This lesson: {SPECIFIC LESSON}. Students: {LEVEL + WHO THEY ARE}. Target length: {5–8} minutes.
Structure (completion-rate engineered):
1. COLD OPEN (20s): the capability they'll have by the end, shown not promised ("By the end you'll be able to look at X and immediately know Y").
2. THE ONE THING: this lesson teaches exactly one core concept. State it, then explain via:
concrete example FIRST → the general principle → a second contrasting example.
3. THE MISTAKE: the error every beginner makes with this, and how to catch it.
4. DO-IT MOMENT (script a pause): a 2-minute exercise the student does NOW, with exact
instructions on screen. Lessons without a do-it moment don't stick.
5. RECAP: 3 bullets, then a 1-line bridge that makes the next lesson feel necessary.
Format as two columns: SPOKEN WORDS | ON SCREEN (slides, screen-recording moments, text overlays).
Tone: energetic teacher, not lecturer — short sentences, direct address, zero filler.

Write a 90-second whiteboard explainer video script for {PRODUCT/SERVICE/CONCEPT} aimed at
{AUDIENCE}.
Format — two columns, synced:
VOICEOVER (left): ~220 words total, conversational. Structure: meet a character with the
audience's exact problem (10s) → the problem gets worse / costs them something real (15s) → enter the solution, explained through the character USING it, not features listed (40s) → the
transformed after-state (15s) → CTA (10s).
DRAWINGS (right): for every voiceover line, what the hand draws — simple nouns and stick-figure actions ("frustrated figure buried under paper stack", "arrow from chaos-cloud to tidy
checklist"). Whiteboard rule: each drawing must be describable in under 8 words or it's too
complex.
Extras: name the character (relatability doubles); one visual metaphor that carries through the whole video; the single sentence viewers should remember.

Help me create a {3 / 4 / 6}-panel comic strip about {TOPIC/JOKE/STORY IDEA}.
Phase 1 — Script the beats: panel-by-panel breakdown — what happens visually in each panel,
dialogue (max 12 words per bubble), and where the beat lands (setup → escalation → punchline in the final panel; the pause-panel before a punchline doubles the laugh).
Phase 2 — Character anchors: 2-sentence visual description per character, specific enough to
repeat verbatim in every panel prompt (this is how they stay consistent).
Phase 3 — Panel prompts: one paste-ready image prompt per panel:
"Comic panel in {STYLE — newspaper strip, webcomic flat color, manga}: [character anchors]. Scene:
[action this panel]. Composition: [close-up/wide, who's positioned where]. Speech bubble space
left {TOP/SIDE}. Clean line art, consistent style, no text in image."
(Text gets added afterwards in Canva — AI mangles bubble text.)
Phase 4 — The lettering guide: exact dialogue per bubble and placement.

Photorealistic {INTERIOR / EXTERIOR} visualization: {SPACE — living room, café, office lobby,
house facade} in {STYLE — japandi, industrial loft, warm minimalism, mid-century modern,
mediterranean}.
Materials palette (name them — this is what sells realism): {e.g. white oak flooring, travertine,brushed brass fixtures, bouclé fabric, matte black steel, linen curtains}.
Layout & key pieces: {WHAT'S IN THE SPACE — furniture, plants, focal element}.
Light: {time of day} with {light quality — soft morning light through sheer curtains / golden hour shafts / overcast even light}, realistic shadows and reflections.
Camera: {wide 24mm room view from corner at eye level / detail vignette of one zone}, straight
verticals (architectural photography standard), f/8 everything sharp.
Mood: {serene / dramatic / cozy / editorial}. Color story: {2–3 COLORS}.
Style reference: as shot for Architectural Digest. Ultra-detailed, no people, no text, no
watermarks.
VARIATIONS: after the first render, request the same room at {night / different material palette / lived-in with everyday objects}.

Professional food photography of {DISH}, styled for {RESTAURANT MENU / INSTAGRAM / COOKBOOK /
DELIVERY APP}.
Plating: served {ON/IN — rustic ceramic plate, cast-iron pan, wooden board}, garnished with {FRESH ELEMENT}, one deliberate imperfection (a crumb, a drip, a torn edge) for realism.
Appetite triggers: {steam rising / glossy sauce sheen / melted cheese pull / fresh herb brightness / charred edges / condensation on glass} — pick what fits the dish.
Scene: {BACKGROUND — dark moody slate, bright marble, linen tablecloth}, supporting props out of focus ({ingredients scattered, cutlery, napkin, drink}).
Camera: {45° angle for most dishes / directly overhead flat-lay for spreads / close macro for
texture}, 85mm, f/2.8, dish fills 60% of frame.
Light: {soft window light from the side / dramatic low-key with deep shadows / bright airy
daylight}. Natural color, appetizing warmth, no oversaturation.
Editorial food magazine quality, ultra-detailed, no text, no hands.

Isometric 3D icon of {OBJECT/CONCEPT}, rendered in a consistent set style: soft clay-like material with subtle matte texture, {PRIMARY COLOR} and {SECONDARY COLOR} palette with one {ACCENT COLOR} highlight, smooth rounded edges, floating on a {SOLID PALE / TRANSPARENT} background with a soft drop shadow beneath. Lighting: gentle top-left studio light, soft ambient occlusion. Camera: true isometric angle (30°), object centered with 20% padding. Style: modern SaaS illustration, playful but professional, no text, no outlines, high detail, 4K.
SET RULE: keep this entire prompt identical for every icon — change ONLY the {OBJECT/CONCEPT}.
Generate these as one batch session so the model stays consistent: {LIST YOUR 6–12 ICONS, e.g.
rocket, shield, graph, envelope, gear, lightbulb}.

Build my 30-60-90 day plan. New role: {TITLE} at {COMPANY TYPE/SIZE}. What I know about their
situation: {ANY CONTEXT — why they hired, team state, known problems}. What success looks like per my manager (if known): {PASTE OR SKIP}.
Structure:
DAYS 1–30 — LEARN: the 10 questions to answer about how things really work (decision-making,
unwritten rules, who influences what); the 5 relationships to build first and what to ask each
person; the map of current processes BEFORE suggesting changes. Include the classic trap to avoid: proposing fixes before earning context.
DAYS 31–60 — CONTRIBUTE: pick my early win — criteria: visible, achievable solo or with light
help, matters to my manager, low political risk. Give 3 candidate types for my role. Start owning routine responsibilities fully.
DAYS 61–90 — LEAD: propose one improvement backed by what I observed (not what worked at my last job); take on something nobody owns; set 6-month goals with my manager.
Finish with: the 5 questions to ask my manager in week 1, and a simple weekly self-check (3
questions).

Rewrite my project description for my {PORTFOLIO / GITHUB README / CV}. Reviewers spend 30 seconds — make them count.
MY PROJECT (rough description): {WHAT IT IS, WHAT IT DOES, TECH USED, ANYTHING HARD ABOUT IT}
Produce:
1. ONE-LINER: what it does + for whom + the impressive bit, in under 20 words. No "simple" or
"basic" — never undersell in the first line.
2. FULL DESCRIPTION (100–140 words): the problem it solves → the interesting decision or trade-off
I made (this signals seniority more than any tech list) → one concrete challenge and how I solved it → outcome (users, performance, or what I learned that changed how I build).
3. TECH LIST: ordered by relevance to the jobs I want ({TARGET ROLE}), not alphabetically.
4. THE QUESTION IT PLANTS: a good project description makes interviewers ask about it. What
question does mine invite, and my 60-second answer.
Rule: specifics beat adjectives. "Handles 10k concurrent users" beats "highly scalable".

Decode this job posting for me. Read it like a skeptical insider, not an applicant.
1. THE REAL JOB: ignoring the wishlist, what will this person spend 80% of their time doing? (The first 3 responsibilities and any repeated theme reveal it.)
2. THE REAL REQUIREMENTS: which listed requirements are hard gates vs. nice-to-haves I can apply without? (Rule of thumb: the top 3 + anything repeated = real.)
3. RED FLAGS: translate the euphemisms — "fast-paced environment", "wear many hats", "like a
family", salary missing, 15 responsibilities for a junior title. Rate overall risk: low medium / high.
4. WHAT THEY'RE STRUGGLING WITH: job ads are complaints in disguise. What problem forced them to open this role?
5. MY ANGLE: given the decoded reality, the 2 things my application should emphasize.
6. INTERVIEW QUESTIONS: 4 sharp questions I should ask THEM, each designed to verify a specific suspicion from above.
JOB POSTING: {PASTE}

Help me explain a gap in my career without apologizing for it. The gap: {WHEN + HOW LONG}. What actually happened: {HONEST REASON — burnout, caregiving, failed business, health, travel, job hunting}. What I did during it (even small things): {ANYTHING — courses, freelance, volunteering, recovery, parenting}. 
Deliver:
1. CV LINE: one honest, neutral line for the CV (e.g. "Career break — family caregiving; completed X during this time"). No euphemisms that invite suspicion.
2. INTERVIEW ANSWER (60 seconds): structure = brief factual reason → what I learned or did → pivot to why I'm ready and motivated NOW. Confident tone; no oversharing, no defensiveness, ends looking forward.
3. THE FOLLOW-UP TRAP: the probing question an interviewer might ask next, and my calm answer.
4. REFRAME CHECK: rewrite anything in my story that sounds like an apology into a statement of
fact.
Rule: never lie, never invent activities — dishonesty in interviews compounds.

I solved a problem and got it wrong (or I'm stuck). Do NOT just show me the correct solution —
diagnose MY attempt first. I learn from my errors, not from your answers.
Process:
1. AUTOPSY: go through my work line by line. Find the exact step where it went wrong. Quote that step back to me.
2. DIAGNOSIS: name the error type — concept misunderstanding, formula misuse, algebra slip, wrong assumption, or misread problem. These need different cures.
3. THE WHY: explain the underlying concept I got wrong, with a tiny example that isolates just
that concept (not the whole problem).
4. GUIDED RETRY: give me a strong hint for fixing my attempt — but let ME redo the step. Wait for my answer.
5. After I retry: confirm or correct, then show the clean full solution and how an expert would think through it from the start ("First I'd notice X, which signals method Y").
6. VACCINE: one similar practice problem targeting exactly my error type, and a one-line rule to remember ("When you see ___, check ___ before ___").
THE PROBLEM: {PASTE}
MY ATTEMPT (including where I got stuck): {PASTE — messy work is fine}

Break down this research paper for me. My background: {FIELD + LEVEL}. Why I'm reading it: {MY
QUESTION / MY PROJECT / EXAM / CURIOSITY}.
Deliver:
1. THE FINDING in one sentence a smart friend would understand.
2. THE SETUP: what they did, with whom/what, measured how — 4 bullets max, no jargon without a
plain-English gloss in parentheses.
3. NUMBERS THAT MATTER: the 3 key results with effect sizes in context ("improved 12% — meaningful because baseline was...") — not just p-values.
4. HOW MUCH TO TRUST IT: sample size adequacy, control quality, who funded it,
correlation-vs-causation status, and whether findings likely generalize beyond the study population. Rate: strong / suggestive / weak evidence.
5. WHAT THE AUTHORS ADMIT: their stated limitations — and one they didn't state.
6. FOR MY QUESTION SPECIFICALLY: what this paper does and does NOT let me conclude about {MY
QUESTION}.
7. READ OR SKIP: which sections of the original are worth my time, and what to read next (cited
papers that matter).
PAPER:
{PASTE TEXT OR UPLOAD PDF}

Write documentation for my code that a stressed developer at 6pm actually finds useful. Audience:
{FUTURE ME / MY TEAM / OPEN-SOURCE USERS}.
Produce:
1. ONE-PARAGRAPH OVERVIEW: what this does and when you'd reach for it — written before any
installation details, because people first decide IF they care.
2. QUICKSTART: the minimum copy-paste path from zero to seeing it work. Every command runnable, every placeholder obvious. If setup has a prerequisite, say it BEFORE the step that fails without it.
3. USAGE EXAMPLES: 3 realistic scenarios (basic, typical, advanced) with actual code — examples teach better than API tables.
4. THE GOTCHAS SECTION (the part everyone forgets to write): surprising behaviors, common
mistakes, error messages users will see and what they actually mean, performance cliffs.
5. API REFERENCE: only for the public surface — parameters, returns, throws. Skip internals.
Rules: no marketing adjectives ("blazingly fast", "simple"), present tense, second person ("you"), and every claim testable.
CODE / PROJECT:
{PASTE}

Run a competitor teardown. My business: {WHAT I DO + WHO FOR}. Competitor: {NAME + URL}. I'll
paste what I can see of theirs below. Analyze in this order:
1. POSITIONING: in one sentence, what do they want to be known for? Who are they clearly NOT
serving? (The gap they leave = my opening.)
2. MESSAGE AUDIT: their headline promises vs. what customers likely actually buy for. Any mismatch is exploitable.
3. PRICING PSYCHOLOGY: what their tiers, anchoring and naming reveal about their target customer and margins.
4. FUNNEL: how a stranger becomes their customer — entry content, lead capture, conversion
trigger. Where does it leak?
5. MOAT CHECK: what would be genuinely hard to copy (brand, network, tech) vs. what just LOOKS
impressive.
6. REVIEWS MINING: from the review snippets I paste — the 3 complaints that repeat. Each one is a positioning gift for me.
7. THE THREE MOVES: (a) one thing to steal shamelessly, (b) one thing to do opposite deliberately, (c) one segment/message they've left open that I should claim this quarter.
THEIR MATERIALS (homepage copy, pricing page, some reviews): {PASTE}

Help me create a lead magnet for {MY BUSINESS/NICHE} targeting {AUDIENCE} who struggle with
{SPECIFIC PROBLEM}.
Phase 1 — The offer: propose 5 lead magnet concepts ranked by perceived value vs. effort. Rules: specific beats broad ("The 7-Email Sequence That Revives Dead Leads" beats "Email Marketing Guide"); a checklist/template/swipe-file usually beats an ebook; solvable-in-one-sitting beats comprehensive.
Phase 2 — After I pick:
- TITLE + SUBTITLE: promise a specific outcome with a timeframe or number
- OUTLINE: 5–8 short sections, each ending with one action the reader takes immediately
- Draft sections 1 and 2 in full (300–400 words each), practical tone, zero filler — respect that they gave me their email for THIS
Phase 3 — The delivery machine:
- OPT-IN PAGE COPY: headline, 3 bullets of what's inside, button text
- DELIVERY EMAIL: subject + 80-word body
- FOLLOW-UP EMAIL (2 days later): checks in, adds one bonus tip, soft-introduces my
{PRODUCT/SERVICE}

Turn my rough story into a LinkedIn post that performs — WITHOUT the LinkedIn cringe.
BANNED: "I'm humbled to announce", fake vulnerability, "Agree?", broetry with one word per line,lessons numbered 1-7 that everyone already knows, and turning a coworker into a prop.
Structure:- LINE 1: the scroll-stopper — start inside the moment of tension or with the surprising outcome. Under 12 words. (First line is 80% of the post's fate; give me 3 options.)
- MIDDLE: the story in 6–10 short lines. Real details beat adjectives — what was said, what it cost, what almost went wrong. One line = one thought. White space is my friend.
- THE TURN: what actually changed or what I actually learned — must be specific enough that it
couldn't appear in anyone else's post.
- ENDING: a genuine question OR a quiet statement. No engagement-bait.
- Length: 120–200 words. Zero hashtags or max 3 niche ones.
MY ROUGH STORY (messy notes fine): {PASTE}
WHO I WANT TO REACH: {AUDIENCE}

Generate 10 AI b-roll video prompts for my content about {TOPIC}. Visual world: {AESTHETIC, e.g.
warm documentary realism / sleek tech minimal / cozy film look}. These clips will play under a
voiceover.
Rules for EVERY prompt (this is what makes AI video work instead of glitch):
- ONE subject + ONE clear motion per shot (AI video fails with complex action)
- Specify camera movement: slow push-in / drift left / static tripod / gentle orbit
- Specify lighting + time of day, consistent with the visual world
- 4–8 second clips, 16:9
- End each prompt with: "smooth motion, photorealistic, no morphing, no text"
Format each as:
SHOT {N} — {what script moment it covers}
PROMPT: {the complete paste-ready generation prompt}
Vary the shots: mix close-up details, medium human moments (hands, silhouettes — avoid clear
faces, AI mangles them), wide establishing scenes, and abstract texture shots. Keep one recurring color motif across all 10 so the edit feels intentional.

Photorealistic portrait photograph: {SUBJECT — age range, style, expression, NOT a real person's name}.
Setting: {ENVIRONMENT, e.g. sunlit artist studio / rainy city street at dusk / neutral studio
seamless}.
Camera: shot on {85mm f/1.4 for classic portrait / 35mm f/2 for environmental / 105mm macro for detail}, {EYE-LEVEL / SLIGHTLY ABOVE / LOW HERO ANGLE}.
Lighting: {Rembrandt lighting with soft key at 45° / golden hour rim light / dramatic
single-source window light / clean beauty lighting with fill}. Catchlights visible in eyes.
Realism anchors: natural skin texture with visible pores, individual hair strands, subtle
asymmetry, fabric weave detail, realistic subsurface skin scattering.
Color grade: {warm editorial / muted film stock / crisp commercial}.
Framing: {HEADSHOT / CHEST-UP / THREE-QUARTER}, shallow depth of field, background softly out of focus.
Avoid: plastic skin, over-smoothing, extra fingers, dead eyes, watermark, text

Design a YouTube thumbnail for my video: "{VIDEO TITLE}" — content: {ONE-SENTENCE SUMMARY}.
Channel style: {DESCRIBE or "clean and modern"}.
Step 1 — The psychology: which click-trigger fits this video best — curiosity gap, transformation (before/after), shock/scale, authority, or drama? Pick ONE and explain in one line.
Step 2 — The layout spec:- FOCAL POINT: what fills 40%+ of the frame (a face with THIS expression / an object / a split comparison)
- TEXT: max 4 words that do NOT repeat the title — they complete it. Give 3 options.
- COLOR: dominant color + one contrast accent, chosen against YouTube's white/dark UI
- The ONE detail that creates intrigue (arrow, circle, glitch, impossible element)
Step 3 — The generation prompt: one paste-ready DALL·E prompt for the background/scene (no text — text gets added in Canva), specifying composition with empty space where the text goes.
Step 4 — The 3-foot test: describe how this thumbnail reads at phone size from 3 feet away. If the message isn't instant, revise.

Act as a brand identity designer. My business: {NAME + WHAT IT DOES}. Audience: {WHO}. Personality (pick words like: trustworthy / playful / premium / technical / warm): {3 WORDS}. Colors I like/hate: {OPTIONAL}.
Phase 1 — Five distinct directions (not five variations of one idea):
1. WORDMARK: typography-led concept
2. ABSTRACT MARK: geometric symbol concept
3. PICTORIAL: literal-but-clever object concept
4. NEGATIVE SPACE: hidden second meaning concept
5. WILDCARD: the unexpected one
For each: the concept in 2 sentences, why it fits my brand words, and one weakness.
Phase 2 — For my chosen direction (or all 5), write a ready-to-paste image prompt:
"Minimalist logo design, {CONCEPT DESCRIPTION}, flat vector style, {COLOR} on white background, clean geometric lines, no text unless specified, no gradients, no 3D, centered, plenty of whitespace, professional brand identity, simple enough to work at 16px favicon size"
Phase 3 — Reality check: remind me that AI logos need a designer's cleanup for trademark safety and vector files, and list the 3 things to check before using one commercially.

Turn my study material below into a complete exam prep kit for {SUBJECT/EXAM}.
Produce:
1. ONE-PAGE SUMMARY: the 20% of concepts that will cover 80% of exam questions, as scannable bullets with key terms bolded.
2. FLASHCARDS: 15 QfiA pairs in "Q: ... / A: ..." format (I'll import to Anki). Front = question or term, back = answer under 25 words.
3. MNEMONICS: memory hooks for the 3 hardest lists/sequences in the material.
4. PRACTICE TEST: 8 questions matching real exam style ({MULTIPLE CHOICE / SHORT ANSWER /
PROBLEMS}), from easy to hard. Answers at the end with 1-line explanations.
5. PREDICTION: the 3 most likely exam questions from this material and the ideal answer skeleton for each.
MATERIAL:
{PASTE NOTES/CHAPTER}

Transform my raw meeting notes/transcript into an action document.
Output exactly this structure:
TL;DR: 3 bullets max — what this meeting decided.
DECISIONS MADE: numbered, with who has final say if mentioned.
ACTION ITEMS: table — Task | Owner | Deadline | (mark [!] if owner or deadline is MISSING so we can fix it).
OPEN QUESTIONS: things discussed but not resolved.
PARKED: raised but deliberately postponed.
FOLLOW-UP EMAIL: ready-to-send summary to participants, under 120 words, action items first.
Flag anything contradictory in the notes. Do not invent owners or dates that weren't stated.
NOTES:
{PASTE}

I'm announcing: {WHAT YOU LAUNCHED/UPDATED — 2–3 sentences with the key benefit}.
Rewrite this natively for 6 channels — each must feel written FOR that platform, not copy-pasted:
1. X/TWITTER: hook tweet + 3-tweet thread. First tweet must work standalone.
2. LINKEDIN: 120–180 words, personal angle ("why I built this"), line breaks for skimming, no hashtag soup (max 3).
3. REDDIT ({SUBREDDIT}): humble, value-first, transparent about being the maker, invites feedback.
Redditors smell marketing — write like a community member.
4. EMAIL to my list: subject (max 45 chars) + 100-word body + single CTA button text.
5. PRODUCT HUNT maker comment: story, what's different, what feedback I want.
6. DISCORD/SLACK community: 2 casual lines + link, like telling friends.

I'm announcing: {WHAT YOU LAUNCHED/UPDATED — 2–3 sentences with the key benefit}.
Rewrite this natively for 6 channels — each must feel written FOR that platform, not copy-pasted:
1. X/TWITTER: hook tweet + 3-tweet thread. First tweet must work standalone.
2. LINKEDIN: 120–180 words, personal angle ("why I built this"), line breaks for skimming, no hashtag soup (max 3).
3. REDDIT ({SUBREDDIT}): humble, value-first, transparent about being the maker, invites feedback.
Redditors smell marketing — write like a community member.
4. EMAIL to my list: subject (max 45 chars) + 100-word body + single CTA button text.
5. PRODUCT HUNT maker comment: story, what's different, what feedback I want.
6. DISCORD/SLACK community: 2 casual lines + link, like telling friends.

Create an SEO content brief for the keyword: "{KEYWORD}".
Include:
1. SEARCH INTENT: what is this searcher actually trying to do/decide? What must the page deliver in the first screen to keep them?
2. TITLE OPTIONS: 3 title tags (max 60 chars) — one listicle, one how-to, one unique angle.
3. OUTLINE: H2/H3 structure covering the topic fully. Mark which sections competitors typically have (must-have) vs. gaps I can win with (differentiator).
4. ENTITIES & TERMS: 12–15 related terms/concepts Google expects on a page about this (not keyword stuffing — topical completeness).
5. FAQ: 4 real questions people ask (People-Also-Ask style) with 40-word answers.
6. META DESCRIPTION: 150 chars, includes keyword, promises the specific value.
7. INTERNAL LINK IDEAS: what related pages should link here and vice versa.

Write 10 ad copy variants for {PRODUCT} targeting {AUDIENCE} on {PLATFORM}. Core benefit: {MAIN
BENEFIT}. Pain point: {PAIN}. Offer: {OFFER/CTA}.
Generate exactly:
- 2 × PAS (Problem fi Agitate -> Solve)
- 2 × Social proof angle (use {PROOF POINT, e.g. "2,000+ users"})
- 2 × Direct benefit ("Get X without Y")
- 1 × Question hook
- 1 × Contrarian/pattern-interrupt ("Stop doing {COMMON PRACTICE}")
- 1 × Story-in-two-lines
- 1 × Us-vs-old-way comparison
Per variant: HEADLINE (max 40 chars), BODY (max {PLATFORM LIMIT} chars), CTA button text. Rate
each 1–5 for risk (safe -> edgy) so I can test a spread. No emoji spam — max 1 per variant.

I want to build: {ONE-SENTENCE APP IDEA}. Target user: {WHO}. I know: {YOUR SKILLS/STACK or "beginner"}.
Give me a shippable plan, not a lecture:
1. SCOPE CUT: the true MVP — 3 features max. List what I should explicitly NOT build in v1 (this list matters most).
2. STACK: one recommendation with a one-line reason, biased toward boring, well-documented tech I can debug. No exotic choices.
3. DATA MODEL: tables/collections with fields and relations, as a simple schema.
4. FILE STRUCTURE: the actual directory tree.
5. BUILD ORDER: 6–10 steps, each completable in one sitting, each ending with something testable.
6. CODE: write step 1 completely, ready to run, with setup commands.

You are a precision tool for {REGEX / SQL / BASH / GIT} one-liners. Two modes:
MODE A — I describe, you build: Give me the exact expression/command, then a piece-by-piece
breakdown table (fragment fi what it does), then 2 test cases showing it working, then the ONE most likely way it fails (edge case) and the hardened version.
MODE B — I paste, you explain: Break it into fragments with plain-English meaning, state what the whole thing accomplishes in one sentence, flag anything dangerous (catastrophic backtracking, full table scan, rm risks), and suggest a more readable alternative if one exists.
MY REQUEST ({A or B}):
{DESCRIBE OR PASTE}

Act as my debugging partner. I have a bug I can't crack. Use systematic diagnosis — do NOT jump to a guessed fix.
Process:
1. First, ask me up to 5 clarifying questions: exact error/behavior, expected behavior, when it last worked, what changed, environment.
2. Form the 3 most likely hypotheses, ranked by probability, each with a 1-line test to
confirm/eliminate it ("add this log line", "run this command").
3. I'll report results; narrow down and repeat until we find the root cause.
4. Then: the fix, WHY it happened (root cause explanation), and one guard (test/assertion/type) that prevents regression.
THE BUG:
{DESCRIBE + PASTE ERROR/CODE}

Help me write a customer case study about {CUSTOMER/PROJECT}.
Phase 1 — Interview me: Ask me exactly 8 questions, one at a time, covering: the customer's
situation before, the specific pain that triggered action, why they chose us over alternatives,
the implementation reality (including friction — honesty sells), the measurable result, and a
quote-worthy emotional moment. Wait for each answer.
Phase 2 — After my answers, write an 800-word case study: narrative structure (situation ->
struggle -> decision -> result), real numbers highlighted, one pull-quote box, subheadings a
skimmer can follow, and zero marketing superlatives — let the facts brag.

Turn my rough notes into a newsletter issue for {AUDIENCE DESCRIPTION}. My newsletter's
personality: {FRIENDLY EXPERT / WITTY CURATOR / NO-NONSENSE ANALYST}.
Format:
- SUBJECT LINES: 3 options — curiosity, benefit, and specific-number style. Max 45 chars each.
- PREVIEW TEXT: 1 line that extends (not repeats) the subject.
- OPENER: 2–3 sentences, warm, references something current or relatable. No "Welcome back!".
- BODY: my items below, each as: bold mini-headline -> 2–3 sentences of context and MY take (not neutral summary) -> link placeholder.
- SIGN-OFF: one useful or fun one-liner + P.S. with a soft ask ({REPLY / SHARE / CHECK THIS OUT}).
Keep total under 500 words — respect inbox time.
MY NOTES:
{PASTE BULLETS}

You are a ruthless line editor at a top publication. Edit my draft below.
Your job:
1. Cut 20–30% of the words without losing meaning — kill redundancies, filler phrases,
throat-clearing intros, and weak qualifiers (very, really, quite, just).
2. Replace passive voice with active. Replace abstract nouns with verbs.
3. Break any sentence over 25 words.
4. Flag (don't fix) with [?]: claims needing evidence, confusing logic, and places where I tell instead of show.
Output: (A) the edited version, (B) a short list of my 3 worst recurring habits with one example each, so I improve.
DRAFT:
{PASTE}

I'll paste a transcript of my YouTube video. Extract the FIVE most viral-worthy short clips.
For each clip give me:
1. START–END quote (the exact transcript lines to cut, 20–45 seconds of speech)
2. HOOK: 1-line on-screen text for the first frame (curiosity gap or bold claim)
3. WHY IT WORKS: one sentence — emotion, surprise, or utility
4. CAPTION: post text with 3 relevant hashtags for {PLATFORM}
5. TITLE: max 40 characters
Prioritize: strong opinions, surprising facts, before/after moments, and complete mini-stories.
Skip anything that needs context from elsewhere in the video.
TRANSCRIPT:
{PASTE}

Write a 60-second talking-head video script about {TOPIC} for {AUDIENCE}, to be delivered by an AI avatar.
Rules: 140–150 words total (natural 60s pace). Structure: relatable question or surprising fact (first 5s) -> the core insight explained with ONE concrete example -> three rapid-fire practical tips -> memorable closing line + CTA to {ACTION}.
Style: spoken language, short sentences, contractions, zero corporate filler.
Add [PAUSE] markers where a beat lands, and [EMPHASIS] on the 3 words that matter most — avatar tools use these for realistic delivery.
End with: a 3-option list of on-screen titles for the video.

Create a complete 30-second vertical video plan for {PRODUCT/SERVICE} targeting {AUDIENCE} on
{TikTok/Reels/Shorts}.
Deliver in this exact format:
HOOK (0–3s): one line that stops scrolling — pattern-interrupt or bold claim, related to {PAIN
POINT}.
SCRIPT: voiceover text, max 65 words total, conversational, no jargon.
SHOT LIST: 6–8 shots with duration, camera direction (close-up/pan/etc.), and what's on screen —
written so an AI video generator OR a phone can execute each shot.
TEXT OVERLAYS: caption for each shot, max 5 words each.
CTA (final 3s): one specific action.
SOUND: suggested music mood + where a sound effect lands.

Using the uploaded photo as the identity reference, re-illustrate this exact scene in {STYLE:
Studio Ghibli-inspired watercolor / 1960s comic book / low-poly 3D / oil painting / pixel art}.
Preserve: the subject's recognizable features, pose, and composition of the original. Transform:
rendering style, color grading to {PALETTE MOOD}, background simplified to {DESCRIPTION}. Add
subtle {STYLE-APPROPRIATE DETAIL, e.g. visible brushstrokes / halftone dots / voxel edges}. Keep
it tasteful — stylized but clearly the same person/scene. High resolution.

Design a {POSTER / EVENT FLYER / BOOK COVER} with the exact text "{YOUR TEXT — keep it under 6
words}" as the dominant headline. Typography: large {BOLD SANS-SERIF / ELEGANT SERIF / RETRO SCRIPT} lettering in {COLOR}, perfectly legible, centered {TOP/MIDDLE}. Visual theme: {DESCRIBE
SCENE OR MOTIF} rendered in {STYLE, e.g. Swiss minimalist / vintage travel poster / synthwave}.
Palette: {2–4 COLORS}. Layout: strong visual hierarchy — headline first, imagery second, small
footer space at bottom for details. Print quality, high contrast, no spelling variations of the
quoted text.

Flat vector illustration in a consistent brand style: {SUBJECT/SCENE}. Style rules: {N}-color limited palette ({LIST HEX CODES}), bold {THICK/THIN} outlines in {COLOR}, simple geometric
shapes, {ROUNDED/SHARP} corners, subtle grain texture, generous negative space. Characters (if
any): diverse, friendly, simplified faces with dot eyes, no photorealism. Background: {SOLID COLOR / SIMPLE SHAPE MOTIF}. Composition: centered subject, breathing room on all sides, suitable as a {BLOG HEADER / APP ILLUSTRATION / SOCIAL POST}. Same style every time — this is one image from a larger brand set.
TIP: Save this prompt. Change only the {SUBJECT/SCENE} each time to keep every image on-brand.

Professional product photography of {PRODUCT}, placed on {SURFACE, e.g. brushed concrete / white marble / walnut wood}, {BACKGROUND, e.g. soft gradient studio backdrop in warm beige}. Lighting: {STYLE, e.g. soft diffused key light from top-left with gentle rim light}, subtle realistic shadows. Shot on 85mm lens, f/2.8, shallow depth of field, product in tack-sharp focus. Composition: {ANGLE, e.g. 3/4 hero angle, product fills 60% of frame}. Mood: {premium / playful /minimal / organic}. Color palette accents: {HEX OR COLOR WORDS}. No text, no watermark, no hands. Ultra-detailed, commercial advertising quality, 4K.

You are a LinkedIn ghostwriter for professionals. Rewrite my profile to attract recruiters for
{TARGET ROLE} positions.
Deliver:
1. HEADLINE (max 220 chars): not just a job title — value + specialty + keyword. Give 3 options:
professional, bold, and keyword-heavy.
2. ABOUT section (max 2,000 chars): first 2 lines must hook (they show before "see more").
Structure: hook -> what I do and for whom -> 3 proof points from my background -> personality line -> call to action. Write in first person, no buzzwords ("passionate", "results-driven" are banned).
3. EXPERIENCE bullets: rewrite the 3 roles I paste, max 4 bullets each, achievement-first.
MY CURRENT PROFILE / CV:
{PASTE}

PROMPT TEXT:
Act as the hiring manager for a {JOB TITLE} position at a {COMPANY TYPE, e.g. mid-size SaaS
company}. Based on my CV below:
1. Write the 12 questions you would most likely ask me — mix of behavioral, technical, and "explain
this gap/choice on your CV" questions specific to MY history.
2. For each, draft a strong 60–90 second answer using ONLY facts from my CV, structured as STAR
(Situation, Task, Action, Result). Where my CV lacks a needed detail, insert {FILL: what happened?} so I know what to add.
3. Finish with the 3 questions I should ask THEM, tailored to this role.
MY CV:
{PASTE CV}

ROLE: You are an expert CV writer and document designer specializing in tech internship applications in Europe (especially Germany). Produce a single-page, ATS-friendly, recruiter-optimized CV in .docx format following the exact specifications below.

🎯 OUTPUT REQUIREMENTS

Format: Single .docx file (A4 paper size, 210 × 297 mm)
Length: Strictly ONE page
Font: Calibri throughout
Margins: Narrow — approximately 1.5 cm top/bottom, 1.6 cm left/right
Language: English (unless told otherwise)
Tone: Professional, confident, achievement-focused, no fluff


🎨 VISUAL DESIGN SPECIFICATION
Color palette (use exactly these hex codes):

Primary accent (NAVY): #1F3A5F — used for name, section headings, company names, links, bullet markers
Body text (DARK): #1A1A1A
Meta/secondary text (GREY): #555555 — used for dates, locations, subtitles

Header layout (centered):

Full name in large NAVY bold caps with wide letter-spacing (~20pt size)
Title line in DARK text directly below — format: [Target Role] | [Degree Program]
Contact line in small GREY text: City, Country • Phone • email@example.com (email in NAVY, underlined)
Links line in small NAVY underlined: github.com/username • portfolio-site.com • other-link.com
A solid NAVY horizontal rule (1.5pt thick) underneath the header to separate it from the body

Section headings:

ALL CAPS, NAVY color, bold, ~11pt with extra letter-spacing
A thin NAVY underline rule beneath each heading
Sections in this exact order: Summary → Professional Experience → Selected Projects → Education → Technical Skills → Languages & Training

Role/job header style (within each section):

Line 1: **Job Title**  •  [Company in NAVY]                              [Dates in italic GREY, right-aligned]
Line 2 (optional subtitle): One italic GREY sentence describing the company/context
Bullets follow

Bullets:

Use a small NAVY square bullet ▪
Each bullet starts with a bold action verb (Developing, Built, Engineered, Implemented, Trained, etc.)
Keep tight line spacing (1.15–1.25) and minimal vertical gap between bullets

Skills section:

Use a 2-column label/value format (not bullets)
Left column: bold NAVY category labels (e.g., "Languages", "ML / Data")
Right column: comma-separated values, with • separators between sub-groups


🧱 STRUCTURE (in this exact order)

Header block (name, title, contact, links, rule)
Summary paragraph — 4–5 lines justified text, no heading needed (or use "PROFESSIONAL SUMMARY"). Mention: current academic status, key skills, current role/highlight, side projects, and what role you're seeking + when.
PROFESSIONAL EXPERIENCE — 1–3 roles, 2–4 bullets each, strongest verbs first
SELECTED PROJECTS — 2–4 projects, 2 bullets each, link the live URL
EDUCATION — Degree, university, dates, focus areas, relevant coursework
TECHNICAL SKILLS — 4–6 rows in label/value format
LANGUAGES & TRAINING — Languages with proficiency levels + certifications combined


✍️ WRITING RULES

Strong action verbs only — start each bullet with: Developing, Engineering, Built, Implemented, Designed, Trained, Founded, Launched, Modeled, Iterating, Documented, etc.
Quantify when possible — but never invent numbers
Be specific with tools — e.g., "Python pipeline with Trimesh and scikit-learn" not "ML pipeline"
Bold key technologies inline within bullets (e.g., **Python**, **scikit-learn**)
No clichés — avoid "passionate", "team player", "go-getter", "results-driven"
No personal pronouns — start with verbs, not "I"
Keep bullets ≤ 2 lines when printed
Dates format: MM/YYYY – MM/YYYY or MM/YYYY – Present


🚫 DO NOT

Add photos, age, marital status, or nationality (unless explicitly required by the user)
Use icons unless requested
Use multiple fonts or decorative graphics
Exceed one page
Include references or "References available upon request"
Use light grey text that won't print well
Use tables with visible borders for the skills section (use invisible alignment)  

<<< YOUR INFO >>>

FULL NAME: [Your full name]
TARGET ROLE: [e.g., "Machine Learning & Data Science Student"]
DEGREE LINE: [e.g., "B.Sc. Information Technology"]
LOCATION: [City, Country]
PHONE: [+xx xxx xxx xxx]
EMAIL: [your.email@example.com]
LINKS: [github.com/yourhandle, portfolio-url.com, etc.]

SUMMARY GOAL:
- Current status: [e.g., "4th-semester B.Sc. student at your universtiy name"]
- Strongest credential: [e.g., "Currently developing ML pipelines for a Fraunhofer-spin-off MedTech project"]
- Notable projects: [e.g., "project name"]
- Target role + start date: [e.g., "Seeking ML/Data Science internship starting Winter Semester 2026/27"]

PROFESSIONAL EXPERIENCE:
[For each role, provide:]
- Job Title:
- Company:
- Dates:
- Location:
- Company subtitle (one line):
- 3–4 achievement bullets (what you did, what tools, what impact):

PROJECTS:
[For each project, provide:]
- Project name:
- Role/type (e.g., "Solo Project", "Founder"):
- URL:
- Dates:
- 2 bullets describing what you built and the tech stack:

EDUCATION:
- Degree:
- Institution:
- Dates + current semester:
- Focus areas:
- Relevant coursework:

TECHNICAL SKILLS (group into 4–6 categories):
- Languages: [e.g., Python, Java, SQL]
- ML / Data:
- 3D / Web:
- Tools:
- (Add or rename categories as needed)

LANGUAGES:
- [Language] ([Level], [Description]) — e.g., English (B2, Full Professional)
- [Language] ([Level]) ...

TRAINING / CERTIFICATES:
- [Certificate name] — [Institution] ([Dates]); [optional notes]

<<< END OF YOUR INFO >>>

You are an expert AI illustrator specializing in character design and sticker art. Using the reference photo provided as the sole visual identity reference, create a sticker sheet that faithfully replicates the exact facial appearance of the subject — including face shape, eye shape and color, nose structure, lip shape, skin tone, eyebrow shape, and hair color, texture, and style. Do not generalize or idealize the features; reproduce them as accurately as possible. Design {STICKER_COUNT} unique sticker panels in {ART_STYLE} style, each showing a different expression including {EXPRESSION_LIST}. The face in every panel must remain consistently identical to the reference photo — only the expression changes, never the facial identity. Below each expression, add a cute label written in {LABEL_LANGUAGE}, for example {LABEL_EXAMPLE}. For the color palette, use only these exact HEX colors throughout the entire illustration — primary color {HEX_1} for skin and base tones, secondary color {HEX_2} for clothing and hair, accent color {HEX_3} for blush and highlights, outline color {HEX_4} for all line art, background color {HEX_5} for each sticker panel, and {HEX_6} for the die-cut outer border. Do not add or substitute any other colors outside of this palette. Each sticker must have consistent line art, a die-cut outline effect, and maintain facial identity consistency across all panels. Output must be high detail, high resolution at least 300 DPI, sticker-ready, and print quality.