brainrot-for-good
Produce high-retention, dopamine-aware video content using brainrot editing techniques — fast cuts, word-by-word captions, sound design, visual velocity, pattern interrupts — all in service of genuine
SKILL.mdskills/video/brainrot-for-good/SKILL.md
GitHubBrainrot for Good — High-Retention Video with Substance
Use the attention-capture techniques of brainrot content — fast cuts, word-by-word captions, sound design, pattern interrupts — to deliver content that actually matters. The pacing of brainrot, the substance of a textbook.
The Ethical Framework
This skill exists for one reason: attention is the scarcest resource, and the techniques that capture it most effectively are currently used for empty content. We reclaim them for value.
The Ethical Test (apply to every piece)
- Value-first: Does every engagement technique serve content that delivers genuine insight?
- Substance survives: If you removed all effects and read the script as plain text, would it still be worth reading?
- Active recall: Does the content prompt action (try this, build this, think about this)?
- No exploitation: Are you capturing attention to deliver value, or just capturing attention?
- Transparency: The audience knows this is engineered for engagement — don't pretend otherwise.
If the script has nothing to teach, no technique will save it. Fix the script first.
Compounding Skills
| Skill | Role |
|---|---|
/launch-video |
Liquid Glass aesthetic, GlassPanel, ParticleField, spring animations |
/remotion-best-practices |
Composition structure, <Sequence>, rendering |
/content-creation |
Storytelling frameworks, AI assets (Imagen, Veo, TTS) |
/google-veo |
Cinematic B-roll generation |
/subtitle-generation |
Word-by-word caption generation |
/blog-post |
Distribution pipeline |
/creative-review |
Style adherence scoring |
The 8 Retention Triggers
Based on dopamine-economy research. Use all 8 in every video:
| # | Trigger | Technique | Ethical Use |
|---|---|---|---|
| 1 | Pattern Interrupt | Visual/audio jolt every 2-4s that resets attention | Emphasize key information, not random noise |
| 2 | Information Gap | Tease the answer before revealing it | Build genuine curiosity about real concepts |
| 3 | Visual Velocity | Scene change every 2-4 seconds | Match cuts to content beats, not arbitrary |
| 4 | Micro-Narrative | Setup → conflict → resolution in 15-60s | Teach through story, not just facts |
| 5 | Sensory Layering | Synced VO + visuals + text + SFX | Each layer reinforces the SAME message |
| 6 | Social Proof | "Most developers don't know this" | Only use when the claim is actually true |
| 7 | Thesis-Antithesis | Present belief → challenge → resolve with insight | Create genuine cognitive engagement |
| 8 | Exit Prevention | Cliffhanger or CTA in final 3 seconds | Point to something worth their time |
Pacing Specs
| Metric | Brainrot Standard | Brainrot-for-Good Target | Fireship Reference |
|---|---|---|---|
| Cuts per minute | 20-30 | 10-15 | 10-15 |
| Words per minute | 180-220 | 200-250 | 200-250 |
| Scene hold time | 1-2s | 2-4s | 2-4s |
| Total duration | 15-60s | 30-120s | 100s (strict) |
| Pattern interrupts | Every 1-2s | Every 3-5s | Every 4-6s |
| Hook window | 1.3s | 3s | 3s |
Key difference: Pure brainrot overstimulates. Brainrot-for-good operates at the cognitive load sweet spot — enough stimulation to hold attention, not so much it overwhelms comprehension.
Caption System (Word-by-Word)
The highest-engagement caption format. Each word appears as it's spoken, with the current word highlighted.
Remotion Implementation
Placement follows content-engine's 9:16 layout contract
(content-engine/references/vertical-layout.md). On a 1080x1920 canvas the caption box
is the caption band: x 220-860, y 1288-1463, centred on 72% of the height and ending
short of the like/comment/share rail (x 872). The old bottom-anchored placement put the
words in the band the Reel UI covers with the username and caption (from y 1686 down).
Keep each group on one line within the band's 640 px, counting the active word's
1.2x scale. At 72 px extra-bold Montserrat that is about 10 uppercase characters, so
shrink the font or split the group. Run the gate with --expect-captions: it fails
captions moved wholly out of the band, but not captions only partly out of it, or
captions elsewhere while other text (your end-card CTA) sits in the band. Check those
on the guide sheet.
// 1080x1920 caption band from content-engine/layout/vertical-9x16.json
const CAPTION_BAND = { left: 220, top: 1288, width: 640, height: 175 };
const WordByWordCaption: React.FC<{
words: string[];
currentWordIndex: number;
}> = ({ words, currentWordIndex }) => (
<div style={{
position: 'absolute',
...CAPTION_BAND,
display: 'flex',
flexWrap: 'wrap',
alignItems: 'center',
alignContent: 'center',
justifyContent: 'center',
gap: 8,
lineHeight: 1.1, // two lines of 72px fit the 175px band; keep groups to one
}}>
{words.map((word, i) => {
const isActive = i === currentWordIndex;
const isPast = i < currentWordIndex;
return (
<span key={i} style={{
fontFamily: 'Montserrat, sans-serif',
fontWeight: 800,
fontSize: 72,
textTransform: 'uppercase',
color: isActive ? '#FFDD00' : isPast ? '#FFFFFF' : 'rgba(255,255,255,0.3)',
transform: isActive ? 'scale(1.2)' : 'scale(1)',
// Stroke, not just a shadow: footage behind brainrot captions is busy.
WebkitTextStroke: '6px #000',
paintOrder: 'stroke fill',
textShadow: isActive
? '0 0 20px rgba(255,221,0,0.5), 2px 2px 0 #000'
: '2px 2px 0 #000',
transition: 'all 0.1s ease',
}}>
{word}
</span>
);
})}
</div>
);
Caption Style Presets
| Style | Font | Active Color | Use Case |
|---|---|---|---|
| Hormozi | Montserrat Bold, uppercase | Yellow (#FFDD00) | Business/productivity |
| Fireship | Inter Bold | White with glow | Technical/dev |
| Clean | SF Pro, sentence case | Blue (#3B82F6) | Professional/launch |
| Chaotic | Comic Sans Bold, random rotation | Rainbow cycling | Meme-aware edutainment |
Sound Design
Sound Effect Library (Ethical Use)
| Sound | When to Use | Effect | Source |
|---|---|---|---|
| Impact/boom | Key stat or revelation | Emphasizes importance | Vine Boom style |
| Whoosh | Transition between scenes | Smooth flow | Standard SFX |
| Cash register | Cost savings, revenue numbers | Monetary emphasis | Standard SFX |
| Notification ping | New concept introduced | Signals attention | UI sounds |
| Typing clicks | Code or terminal scenes | Authenticity | Keyboard ASMR |
| Bass drop | Plot twist or contrarian reveal | Dramatic emphasis | Music production |
Rule: Sound effects emphasize key information. Never use them as empty stimulation. Every sound must correspond to a content beat.
Background Music
- Lofi beats (90-110 BPM) for tutorials
- Electronic/synthwave (120-140 BPM) for product demos
- No vocals — compete with narration
- Volume: -18dB below narration
Scene Types (Brainrot Variants of /launch-video)
1. Rapid-Fire Intro (replaces Title Card)
Duration: 3 seconds Content: 3 fast cuts — problem image → solution image → result stat Audio: Three ascending impact sounds Caption: "STOP. SCROLLING." → topic in 5 words
2. Terminal Speed Run (replaces Prompt Scene)
Duration: 4-6 seconds Content: Terminal with FAST typewriter (3-4 chars/frame), code flying by Audio: Rapid keyboard ASMR, subtle bass build Caption: Word-by-word narration of what's happening
3. Pipeline Blitz (replaces Pipeline Scene)
Duration: 4-5 seconds Content: All 9 nodes SLAM in simultaneously with stagger of 3 frames each Audio: Machine-gun impact sounds, ascending pitch Caption: "NINE. PHASES." (Hormozi style, big text)
4. File Explosion (replaces Product Demo)
Duration: 3-4 seconds Content: Files spray outward from center like an explosion, then snap into a grid Audio: Whoosh → snap into place sound Caption: "18 files. One command."
5. Checkmark Cascade (replaces Platform Success)
Duration: 3 seconds Content: Platform cards slam in from edges, checkmarks EXPLODE with particle burst Audio: Four rapid success chimes, ascending Caption: "X. LinkedIn. Instagram. Blog. DONE."
6. CTA Slam (replaces CTA Card)
Duration: 3 seconds Content: Command text SLAMS onto screen with bounce animation Audio: Final impact + brief silence Caption: Word-by-word "install it right now"
Composition Structure
Total: 20-30 seconds (brainrot-optimized)
[0-3s] Rapid-Fire Intro — 3 fast cuts with impacts
[3-7s] Terminal Speed Run — fast typewriter in glass panel
[7-11s] Pipeline Blitz — 9 nodes slam in rapid sequence
[11-15s] File Explosion — outputs spray and snap to grid
[15-18s] Checkmark Cascade — platforms verified with particle bursts
[18-21s] CTA Slam — install command with bounce
Background: ParticleField (doubled particle count, faster drift)
Captions: Word-by-word throughout, Hormozi style
Music: Synthwave 130BPM, -18dB
SFX: Impact on every scene transition
Production Pipeline
1. SCRIPT → Write narration first (200-250 WPM, every word must teach)
→ Apply ethical test: does substance survive without effects?
2. STORYBOARD → Map each sentence to a scene with visual + SFX
3. ASSETS → Imagen 4.0 for key frames, Veo 3.1 for 2-3s B-roll bursts
4. COMPOSE → Remotion composition with:
→ WordByWordCaption component synced to narration timing
→ SFX timed to content beats (not arbitrary)
→ Visual velocity: scene change every 2-4 seconds
→ ParticleField with 80+ particles (doubled from launch-video)
5. RENDER → npx remotion render BrainrotComposition --output video.mp4
6. REVIEW → /creative-review with brainrot-specific checklist
7. DISTRIBUTE → /blog-post publish pipeline
Quality Checklist (Brainrot-for-Good Specific)
Must Pass
- Script has genuine substance (ethical test: plain text is still worth reading)
- Hook captures in first 3 seconds (pattern interrupt, not just text)
- Word-by-word captions present and synced
- 9:16 layout gate meets the pass criteria in content-engine SKILL.md "Vertical Layout Gate":
python3 ~/.claude/skills/content-engine/scripts/check_vertical_layout.py video video.mp4 --expect-captions - Sound effect on every scene transition
- No static shot longer than 4 seconds
- Total duration 20-60 seconds
- CTA in final 3 seconds
Should Pass
- At least 1 thesis-antithesis moment (challenge a belief)
- At least 1 concrete number or stat
- Background music present, below -18dB
- Pacing at 10-15 cuts per minute
- Active recall prompt ("try this" or "think about this")
Must NOT
- No engagement technique without corresponding content value
- No sensory stimulation that competes with comprehension
- No misleading social proof ("99% don't know this" — if untrue)
- No infinite-scroll optimization (video must have a clear endpoint)
- No empty split-screen gameplay (if split-screen, both halves must serve the message)