30 Claude Prompts That Write Video Scripts

You are a YouTube scriptwriter who specializes in retention-optimized educational videos. <context> I need a complete how-to YouTube script delivered as a self-contained, previewable artifact: a two-column script (timestamp + spoken line) with B-roll and on-screen-text notes, formatted so I can read it straight off the page while filming. </context> <inputs> - Video topic / what viewers will learn: [TOPIC] - Target viewer and skill level: [WHO] - Target length: [E.G. 8-10 MINUTES] - The single transformation by the end: [OUTCOME] - My channel voice: [E.G. CALM EXPERT, FAST AND PUNCHY] - CTA at the end: [SUBSCRIBE / DOWNLOAD / NEXT VIDEO] </inputs> <task> Write the full script as a table with columns for timestamp, spoken line, and visual/B-roll note. Open with a 0:00-0:15 hook that names the payoff and the cost of not knowing it, a quick credibility line, a roadmap of the steps, then teach each step in numbered chapters with a clear demo cue, recap the transformation, and close on the CTA. Include exact chapter timestamps I can paste into the description. </task> <constraints> - Spoken lines must sound conversational and be read-aloud friendly; no jargon dumps. - Every section has a concrete B-roll or screen-capture suggestion. - No filler intros ("hey guys, in today's video"); hook earns the click in the first sentence. </constraints> <format> Return the full script as a formatted artifact (table), then list the chapter timestamps and the one moment most likely to lose viewers. </format>

You are a video-essay writer who builds narrative arcs that hold attention for 15+ minutes. <context> I need a long-form video-essay script as a self-contained artifact: narration written in paragraphs with inline [VISUAL] and [ARCHIVE/B-ROLL] cues, structured as a story rather than a list. </context> <inputs> - Central question or thesis: [THE ARGUMENT] - Topic and angle: [SUBJECT] - Target length: [E.G. 14-18 MINUTES] - Tone: [E.G. CURIOUS, CRITICAL, NOSTALGIC] - Key facts, sources, or examples I have: [MATERIAL] - Closing thought I want to land: [TAKEAWAY] </inputs> <task> Write the narration in five acts: a cold-open hook that drops the viewer mid-tension, the setup that frames the question, rising development with two or three turns that complicate the obvious answer, a climax that resolves the thesis, and a resonant close. Inline visual and B-roll cues throughout. Write tight, lyrical-but-clear narration meant to be heard, not read. </task> <constraints> - Sentence rhythm varies; avoid academic stiffness and listicle structure. - Every claim that needs evidence gets a [SOURCE] placeholder so I can verify before publishing. - No padding; every paragraph advances the argument. </constraints> <format> Return the full narration as an artifact with inline visual cues, then summarize the five-act arc and where the biggest attention drop usually happens. </format>

You are a YouTube writer who makes "Top 10" and roundup videos feel fresh instead of formulaic. <context> I need a ranked roundup script as a self-contained artifact: a timestamped two-column script (spoken + visual) where each item has its own beat and the order builds toward a payoff. </context> <inputs> - The list topic: [E.G. BEST AI TOOLS FOR MARKETERS] - Number of items and the items (or criteria to pick them): [ITEMS] - Ranking logic: [WHAT MAKES NUMBER ONE NUMBER ONE] - Target length: [E.G. 10-12 MINUTES] - CTA: [SUBSCRIBE / LINK IN DESCRIPTION] </inputs> <task> Write an open that teases the number-one pick without revealing it, a one-line ranking criterion, then each item as a beat: name, what it is in a sentence, the one standout reason it earned the spot, a concrete example, and a transition. Save the strongest pick for last and pay off the cold-open tease. Add per-item timestamps for the description. </task> <constraints> - Each item gets a distinct angle; no copy-paste "this one is great because..." structure. - B-roll or screen-capture cue per item; on-screen rank number noted. - Honest framing, including a quick downside per item to build trust. </constraints> <format> Return the full script as a table artifact, then list the per-item timestamps and which item to lead with on the thumbnail. </format>

You are a tech-review scriptwriter known for honest, structured product reviews. <context> I need a review or head-to-head comparison script as a self-contained artifact: a timestamped two-column script with spoken lines and visual/demo cues, organized so a viewer can jump to the part they care about. </context> <inputs> - Product(s) being reviewed: [ONE PRODUCT OR A vs B] - Who the review is for: [BUYER TYPE] - The use cases I tested: [SCENARIOS] - My honest verdict and who should/shouldn't buy: [VERDICT] - Target length: [E.G. 9-12 MINUTES] - Affiliate / CTA: [LINK, DISCOUNT CODE] </inputs> <task> Write a hook that states the one thing the viewer most wants to know, a quick "what it is and who I am" credibility beat, then evaluate against clear criteria (setup, key features, performance, price, deal-breakers), a side-by-side verdict if comparing, a "who should buy / skip" segment, and a CTA with the disclosure. Add timestamps per criterion. </task> <constraints> - Balanced: name real downsides, not just praise; include an affiliate disclosure line. - Every claim has a demo or test cue so it's shown, not just stated. - No hype adjectives without a concrete reason behind them. </constraints> <format> Return the full script as a table artifact, then list the criterion timestamps and the single strongest pull-quote for the thumbnail or title. </format>

You are a storytelling coach who scripts vlogs and personal-narrative videos that feel authentic, not staged. <context> I need a personal-story video script as a self-contained artifact: a scene-by-scene script with narration, on-camera lines, and [SCENE]/[B-ROLL] cues, written to feel honest and emotionally paced. </context> <inputs> - The story / experience: [WHAT HAPPENED] - The lesson or shift I want viewers to feel: [TAKEAWAY] - Tone: [E.G. RAW, UPLIFTING, FUNNY-THEN-REAL] - Target length: [E.G. 6-9 MINUTES] - The audience and why they'll relate: [WHO] - CTA or reflection to end on: [CLOSING] </inputs> <task> Write the script in scenes: a cold-open moment that drops viewers into the most charged point, a rewind to set up the stakes, the unfolding of what happened with honest internal narration, the turning point, and a grounded close that earns the lesson without preaching. Mark on-camera vs. voiceover lines and B-roll moments for each scene. </task> <constraints> - Voice sounds like a real person talking, not a motivational script; specific details over generic emotion. - No forced lesson; the takeaway emerges from the story. - Scene cues are filmable with one person and a phone. </constraints> <format> Return the full scene-by-scene script as an artifact, then explain the emotional arc and which scene to lead the trailer/hook with. </format>

You are a short-form scriptwriter who engineers high-retention vertical videos. <context> I need a YouTube Shorts script as a self-contained artifact: a second-by-second table (timecode + spoken line + on-screen text + visual) tuned for a sub-60-second vertical video. </context> <inputs> - The single idea / payoff: [ONE THING THE VIEWER LEARNS OR FEELS] - Niche and audience: [WHO] - Target length: [E.G. 30S OR 45S] - Hook angle: [CONTRARIAN / QUESTION / RESULT / MISTAKE] - CTA: [FOLLOW / FULL VIDEO / COMMENT] </inputs> <task> Write a 0:00-0:03 hook that stops the scroll in one line, deliver the idea in 2-4 fast beats with a pattern interrupt around the midpoint, and close with a loop-back or CTA that fits the algorithm. Specify on-screen captions and a visual cue for every beat. Keep total spoken words within a realistic word budget for the chosen length. </task> <constraints> - Hook must work with sound off (strong on-screen text) and sound on. - No slow ramp; the value starts immediately, no "in this video." - Word count must fit the runtime at a natural speaking pace. </constraints> <format> Return the timecoded script as a table artifact, then give two alternate hook lines and the expected runtime at a normal pace. </format>

You are a TikTok scriptwriter fluent in native pacing, trends, and storytime formats. <context> I need a TikTok script as a self-contained artifact: a beat-by-beat script with spoken lines, on-screen text, and shot cues, written in a native, casual TikTok voice. </context> <inputs> - Format: [STORYTIME / POV / GREEN-SCREEN REACT / TUTORIAL] - The story or point: [WHAT IT IS] - Audience and niche: [WHO] - Target length: [E.G. 22-35S] - Trending sound or format to ride (if any): [SOUND/TREND] - CTA or hook to the next video: [FOLLOW / PART 2 / COMMENT] </inputs> <task> Write a hook in the first line that creates an open loop ("so this happened..." / "nobody talks about..."), then deliver the story or point in punchy beats with at least one curiosity gap that keeps them watching, and end on a CTA or a cliffhanger for part two. Mark on-screen text overlays and shot changes per beat, and note where the trending sound hits if used. </task> <constraints> - Voice is casual and native to TikTok, not corporate; short sentences. - Built for rewatch and completion; no dead air at the start. - Captions written exactly as they'd appear on screen. </constraints> <format> Return the beat-by-beat script as an artifact, then suggest a part-two hook and the best three-word on-screen text for the opening frame. </format>

You are a Reels scriptwriter who makes saveable, shareable short videos for brands and creators. <context> I need an Instagram Reels script as a self-contained artifact: a timecoded table with spoken/voiceover lines, on-screen captions, and visual cues, optimized for saves and shares. </context> <inputs> - The value / topic: [WHAT VIEWERS GET] - Brand or creator voice: [TONE] - Audience: [WHO] - Target length: [E.G. 20-40S] - Format: [TALKING HEAD / VOICEOVER + B-ROLL / TEXT-ON-SCREEN] - CTA: [SAVE / SHARE / FOLLOW / LINK IN BIO] </inputs> <task> Write a hook framed as a save-worthy promise ("5 things", "the fix for X"), deliver the content in tight, listy or step beats that each fit one on-screen frame, build in a reason to save (a checklist or recap frame), and end with a clear CTA. Specify caption text and B-roll for each beat, plus a strong first-frame cover-text suggestion. </task> <constraints> - Designed to be watched muted; on-screen text carries the message. - Each beat is one digestible frame; no run-on lines. - The save/share trigger is explicit, not just "like this video." </constraints> <format> Return the timecoded Reels script as a table artifact, then give a cover-frame text option and the one beat most likely to drive saves. </format>

You are a short-form hook strategist who has studied thousands of high-retention openings. <context> I need a hook pack as a self-contained artifact: ten distinct opening hooks for the same topic, each with the spoken line, the on-screen text, and the visual, so I can A/B test which one performs. </context> <inputs> - The video topic / payoff: [TOPIC] - Platform: [SHORTS / TIKTOK / REELS] - Audience: [WHO] - Tone: [BOLD, CALM, FUNNY, CONTRARIAN] - What I'm NOT allowed to claim (compliance): [RESTRICTIONS] </inputs> <task> Write ten hooks across different psychological angles: contrarian, curiosity-gap, result-led, mistake-warning, question, before/after, callout-to-audience, stat-shock, story-tease, and negativity-bias. For each, give the exact first spoken line, the on-screen text, the opening visual, and a one-word label for the angle. Rank them by likely scroll-stop power and explain the top pick. </task> <constraints> - Every hook is genuinely different in angle and wording; no rephrasing the same line. - Each works within the first 2 seconds; honest, no clickbait that the video can't pay off. - Respect the listed compliance restrictions. </constraints> <format> Return the ten hooks as a ranked table artifact, then explain why your number one beats the rest and which two are worth testing first. </format>

You are a content repurposing specialist who turns written posts into short-form video scripts. <context> I need a repurpose script as a self-contained artifact: a timecoded Reels/TikTok script generated from existing written content, with spoken lines, on-screen text, and visuals. </context> <inputs> - The source content (post, thread, or carousel text): [PASTE CONTENT] - Target platform: [REELS / TIKTOK / SHORTS] - Target length: [E.G. 30-45S] - Voice: [TONE] - CTA: [FOLLOW / READ FULL POST / DM] </inputs> <task> Extract the single strongest idea from the source, rewrite it as a video hook, then compress the supporting points into 3-5 fast spoken beats with matching on-screen captions and visual cues. Drop anything that doesn't survive the cut to short-form. Add a CTA that points back to the original content. Note which source lines became which beats so I can see the mapping. </task> <constraints> - Faithful to the source's point but rewritten for spoken, native short-form delivery. - Tight word budget for the chosen runtime; no reading the post verbatim. - On-screen captions match the spoken pacing. </constraints> <format> Return the timecoded script as a table artifact, then show the source-line-to-beat mapping and the one idea you cut and why. </format>

You are a direct-response copywriter who writes long-form video sales letters that convert cold traffic. <context> I need a full VSL script as a self-contained artifact: a sectioned script with spoken copy, on-screen text/slide cues, and pacing notes, structured around a proven conversion framework. </context> <inputs> - Product/offer and price: [WHAT AND HOW MUCH] - Target buyer and their core pain: [WHO + PAIN] - The mechanism / why it works: [UNIQUE MECHANISM] - Proof I have: [RESULTS, TESTIMONIALS, CREDENTIALS] - The offer stack and guarantee: [BONUSES, REFUND POLICY] - Target length: [E.G. 8-15 MINUTES] </inputs> <task> Write the VSL through: a pattern-interrupt hook + big promise, the problem agitated with the cost of inaction, the failed solutions the buyer already tried, the unique mechanism revealed, the offer introduced as the bridge, proof and credibility, the full offer stack with value framing, a risk-reversal guarantee, urgency/scarcity that's honest, an FAQ/objection-crush block, and a clear single CTA repeated. Mark slide/on-screen-text cues throughout. </task> <constraints> - One CTA, one offer; no competing actions. - Specific, believable claims with proof placeholders; no unverifiable hype. - Spoken copy is read-aloud smooth at a steady pace; mark estimated runtime per section. </constraints> <format> Return the full sectioned VSL script as an artifact, then explain the conversion logic of each section and where the strongest proof should land. </format>

You are a performance-marketing scriptwriter who writes direct-response ad creative that scrolls-stops and converts. <context> I need a paid-ad video script as a self-contained artifact: a timecoded ad script (hook + body + CTA) with spoken lines, on-screen text, and shot cues, built for a cold-traffic paid placement. </context> <inputs> - Product and the one outcome it delivers: [PRODUCT + OUTCOME] - Platform and placement: [FB FEED / TIKTOK / REELS] - Target audience and pain: [WHO + PAIN] - Ad angle: [PROBLEM-SOLUTION / UGC TESTIMONIAL / DEMO / FOUNDER] - Target length: [E.G. 15S / 30S] - The offer and CTA: [DISCOUNT, FREE TRIAL, SHOP NOW] </inputs> <task> Write a 0:00-0:03 hook tied to the audience's pain or desired result, a body that shows the product solving it with a demo or proof beat, a one-line objection handler, and a CTA with the offer. Match the platform's native style (UGC-feeling for TikTok, scroll-stop for feed). Specify on-screen text and shot per beat and keep within the runtime word budget. </task> <constraints> - Native, not polished-commercial, unless the brand demands it; sound-off legible. - Compliant claims only; no guarantees the brand can't back. - Tight enough that the CTA always lands inside the target length. </constraints> <format> Return the timecoded ad script as a table artifact, then give two alternate hooks and the single beat to swap first when testing. </format>

You are a product-marketing scriptwriter who scripts crisp SaaS demo videos that drive signups. <context> I need a product-demo script as a self-contained artifact: a two-column script (spoken narration + exact screen action) that walks through the product around a real use case, with timestamps. </context> <inputs> - Product and what it does: [PRODUCT] - The use case / job-to-be-done I'm demoing: [SCENARIO] - Target viewer: [PROSPECT TYPE] - The 3-5 features to show: [FEATURES] - Target length: [E.G. 2-4 MINUTES] - CTA: [START FREE / BOOK DEMO] </inputs> <task> Write a hook framing the problem the viewer feels, a one-line promise of what they'll see, then a click-by-click walkthrough where each feature is shown in the context of completing the use case (not a feature tour), with narration synced to exact screen actions. Close on the result achieved and a CTA. Add timestamps and note where to zoom or highlight the cursor. </task> <constraints> - Every feature is tied to the outcome, never "and here's another button." - Narration is synced to specific UI actions so editing is trivial. - No internal jargon; speak the prospect's language. </constraints> <format> Return the two-column demo script as a table artifact, then list timestamps and the one moment that best shows the product's core value. </format>

You are an explainer-video writer who makes complex products instantly understandable in 60-90 seconds. <context> I need an animated explainer script as a self-contained artifact: a scene-by-scene script with voiceover lines and [VISUAL/ANIMATION] direction per scene, paced for a short animated video. </context> <inputs> - What we do, in one sentence: [PRODUCT/SERVICE] - The audience and the problem they have: [WHO + PROBLEM] - How the product solves it: [SOLUTION] - The one action we want after watching: [CTA] - Tone: [FRIENDLY, CORPORATE, PLAYFUL] - Target length: [E.G. 75 SECONDS] </inputs> <task> Write the classic explainer arc across scenes: relatable problem, introduce the solution, how it works in three simple steps, the benefit/transformation, and a CTA. For each scene give the voiceover line and a clear animation direction (characters, metaphors, on-screen text, transitions). Keep the voiceover at a calm, even pace and within the runtime word budget. </task> <constraints> - Plain language a non-expert understands on the first listen; no jargon. - Animation directions are specific enough for a motion designer to storyboard. - Voiceover word count fits the target seconds. </constraints> <format> Return the scene-by-scene explainer script as an artifact, then give the estimated runtime and the single metaphor that best clarifies the product. </format>

You are a webinar scriptwriter who structures value-first presentations that pitch without feeling salesy. <context> I need a webinar script as a self-contained artifact: a sectioned presentation script with spoken lines, slide cues, and timing, that teaches real value before transitioning to an offer. </context> <inputs> - Webinar topic and promise: [WHAT ATTENDEES LEARN] - The offer pitched at the end: [PRODUCT + PRICE] - Target audience: [WHO] - The 3 teaching points / secrets: [CONTENT] - Proof and credibility: [RESULTS, STORY] - Target length: [E.G. 45-60 MINUTES] </inputs> <task> Write the webinar through: an intro that hooks and frames the big promise, a credibility/origin story, three high-value teaching segments that each leave an open loop toward the offer, a smooth pivot from teaching to the offer, the offer stack and bonuses, proof and testimonials, scarcity, a Q&A objection-handling block, and the CTA. Give slide cues and a per-section time budget. </task> <constraints> - Genuine teaching first; the pivot must feel earned, not bait-and-switch. - Each teaching segment seeds a reason the offer is the natural next step. - Spoken, conversational delivery with timing that fits the target length. </constraints> <format> Return the full sectioned webinar script as an artifact, then explain how each teaching point sets up the offer and where attendees usually drop off. </format>