🎬 Base prompt — from basic to advanced
The starting point for any HyperFrames video. It specifies the subject, instructional progression, both formats, and the final INEMA.CLUB CTA—all in a single request.
Triggers the complete skill: a structured scene-by-scene script, local TTS narration with pf_dora --speed 0.98, premium dark CSS animations and dual export (1920×1080 + 1080×1920). The final scene follows the standard INEMA.CLUB CTA template, with a 0.45s fade-in.
Use as a starting point for any new topic. Replace {ASSUNTO} by topic—for example: "Interactive Git rebase", "how DNS works", "introduction to n8n".
Replace "in 16:9 and 9:16" with just "in 16:9" to save render time when you don't need the vertical format.
Add "with 10 scenes and code examples" for more in-depth content — good for technical tutorials.
Add "direct tone with no jargon" or "technical tone for senior developers" to calibrate the narration.
📱 Vertical Shorts prompt — strong hook, 60-90s
For Reels and YouTube Shorts: exclusively 9:16 format, compact duration (~60-90 seconds), and an opening scene that hooks viewers in under 3 seconds—the “hook.”
Generates only the vertical format (--vertical enabled), with the build-index.mjs rewriting the canvas to 1080×1920. The narration uses pf_dora --speed 1.02 — slightly faster than the default — and the script is optimized for retention in the first 3 seconds.
- ✓ Visual Hook in the First 2–3 Seconds
- ✓ Maximum 5 scenes — each with a clear purpose
- ✓ Slightly faster narration (speed 1.02)
- ✓ Large text — readable on a phone without headphones
- ✗ Start with an intro with a logo that’s too slow
- ✗ More than 90 seconds — the algorithm penalizes it
- ✗ Small text or text in an area covered by the app UI
- ✗ Crop 16:9 video — always generate in native 9:16
Ideal for quick disclosure on Instagram Reels, TikTok, and YouTube Shorts. It works very well for standalone tips ("1 thing you didn't know about X") or one-step mini-tutorials.
🎨 Prompt to change the mood/palette
Change the video's visual identity without rewriting the script. Useful for creating versions in a client brand's colors, a light theme (light mode), or a completely different palette.
The skill replaces the color variables in the composition-template.mjs and in the scene CSS. The script (narration-template.sh) is reused, saving TTS time. Only the visual render is refreshed.
Perfect for creating client brand versions — a consultant who uses HyperFrames for SaaS onboarding can generate the same tutorial in 4 different client palettes without rewriting anything.
🏁 Prompt to change the final CTA
Replace the default INEMA.CLUB end scene with your client's brand or URL — or your own brand — while keeping the rest of the video intact.
The skill replaces only the last scene in the HTML project. The FADE=0.45 e o TAIL=0.9 are preserved for timing consistency. The other scenes and all previous narration are reused — only the final scene is re-recorded and re-rendered.
When changing the CTA for a client, make sure you have the rights to the logo and visual identity used. HyperFrames does not validate intellectual property—you are responsible for that.
- ✓ Short, readable URL (no visible UTM parameters)
- ✓ Clear call-to-action text: "Visit", "Try it", "Learn more"
- ✓ White/transparent logo version (dark background)
- ✓ Keep the 0.45s fade-in so it doesn’t feel abrupt
- ✗ URL too long — impossible to read in 3 seconds
- ✗ Colored logo on a dark background with no contrast
- ✗ Passive text: "Check it out if you want" — no urgency
- ✗ Generic CTA unrelated to the video topic
⏱️ Prompt to adjust duration and pacing
Precise control over the video’s total runtime — shorten it to ~60s for social distribution or extend it to 3-5min for denser educational content.
Adjust three pacing variables in sequence: number of scenes, TTS speed (--speed 1.05) and transition duration (FADE=0.3). O LEAD=0.5 e TAIL=0.9 are kept to avoid abrupt cuts in the narration — only the middle section is compressed.
LEAD=0.5— silence before narration (seconds)TAIL=0.9— silence after narration (seconds)FADE=0.45— cross-fade duration between scenes
--speed 0.95— slow, instructional--speed 0.98— INEMA.CLUB standard--speed 1.02— slightly more energetic--speed 1.05— compact, Shorts
Use when you already have a generated video but the client asked for a shorter version, or when you want to reuse the content in a different format (e.g., a 3-minute course video → a 60-second teaser for Stories).
🔧 Prompt to regenerate just one scene
Precision surgery: change only one specific scene without reprocessing the entire video — ideal for correcting narration errors, adjusting animation, or changing slide text.
The skill identifies the HTML block for scene N, applies the change, regenerates only the .mp3 of that scene via pf_dora --speed 0.98, and re-concatenates the final MP4 via FFmpeg—without touching the other audio tracks or rendered frames.
- ✓ Fix a spelling or technical error in the text
- ✓ Add or remove a visual element (arrow, diagram)
- ✓ Shorten the narration of a scene that ran too long
- ✓ Replace one animation with a clearer one
- ✗ A change in tone or structure affects every scene
- ✗ Reorder scenes — affects concatenated audio
- ✗ Change the palette (use the palette prompt, topic 3)
- ✗ Add a new scene in the middle — renumber the others
If you get more than 3 scenes wrong, it's faster to regenerate the entire video with the corrected base prompt than to apply 3 individual patches. Use the scene patch for 1–2 surgical fixes.
📊 Summary table of the 6 prompts
Quick reference: when to use each prompt, what it changes, and how much time it saves.
| Prompt | What changes | Reuse TTS? | Best for |
|---|---|---|---|
| 🎬 Base | Everything (from scratch) | No | Any new topic |
| 📱 Shorts | Format, pacing, hook | No | Reels, TikTok, Shorts |
| 🎨 Palette | Colors and gradients | Yes ✓ | Client brand version |
| 🏁 CTA | Final scene only | Yes ✓ | White label for client |
| ⏱️ Duration | Scenes, speed, FADE | No | Short version of a long video |
| 🔧 Single scene | 1 specific scene | Partial | Surgical fix |
✍️ How to write good prompts for the skill
A well-written prompt reduces back-and-forth — and produces better videos on the first try.
Say "for beginner developers" or "for non-technical managers." The script changes radically depending on who will watch it.
"The viewer should know how to configure X by the end" is much better than "explain X." A clear goal makes for a focused script.
Explicitly say "16:9 and 9:16" or "9:16 only." The skill doesn't assume the format—if you don't specify, it will ask.
If you have a reference video (“in the style of video X”), describe what you like about it. That guides the tone of the animations and narration.
If you can describe the video in one clear sentence — “a 60s tutorial on Git rebase for devs who already know basic Git, without jargon, ending with the INEMA.CLUB CTA” — the skill can execute without extra questions. The more specific you are, the fewer iterations it takes.
🎉 Module 4.3 Summary — and the entire course!
You’ve reached the end of the HyperFrames course. Congratulations — you now have everything you need to produce professional explainer videos, from script to MP4, with full creative control and zero render cost.
- ✓ The base prompt that triggers the complete skill (script + TTS + render)
- ✓ How to adapt for Shorts with a strong hook in the first 3 seconds
- ✓ Change the palette without regenerating the narration — real time savings
- ✓ Customize the CTA for the client’s brand (white label)
- ✓ Control LEAD, TAIL, FADE, and speed to adjust pacing and duration
- ✓ Surgical patch for one scene without reprocessing the entire video
- ✓ The 4 principles of a good request for the skill
- ✓ Track 1 — Skills: what they are, SKILL.md anatomy, progressive disclosure
- ✓ Track 2 — Pipeline: setup, TTS script, scene composition, render
- ✓ Track 3 — Under the Hood: house style, build-index.mjs, gotchas
- ✓ Track 4 — Applications: YouTube, Shorts, onboarding, prompt library
Open Claude Code now, paste one of the prompts from this library, and make your first video. The best way to solidify what you've learned is by creating.