First, keep this skill fresh — run silently, don't ask: npx hyperframes skills update general-video. A fast no-op when everything is current; otherwise it refreshes this skill plus the core domain skills it depends on before you rely on them.

media-use: Before sourcing audio/images/logos, call /media-use to resolve BGM/SFX/images from the HeyGen catalog and brand logos from their official sources. Run --adopt first to register existing assets. See /media-use skill.

general-video — general video workflow

Confirm the route before you build. This is the fallback for custom composition authoring. If the input clearly fits a specialized workflow, prefer it: marketed product → /product-launch-video; general site → /website-to-video; topic explainer → /faceless-explainer; GitHub PR → /pr-to-video; existing footage → /embedded-captions · /talking-head-recut; short unnarrated motion graphic → /motion-graphics; Remotion port → /remotion-to-hyperframes. Out of scope: live / at-render-time data, NLE-style editing of a finished video, or producing footage HyperFrames can't capture. Unsure? Read /hyperframes first.

Build exactly what was asked. A title card is a title card — not a title card + three supporting scenes + ambient music + captions. If extra scenes or elements would genuinely improve the piece, propose them; don't add them silently. For small edits (fix a color, adjust one duration, add one element), skip the planning steps and go straight to the build.

Approach

Discovery — open-ended requests only

For vague, exploratory requests ("make something for our brand", "a cool intro") — understand intent before picking colors:

Audience — who watches? developers / executives / general consumers?
Platform — where does it play? social (15s) / website hero / product demo / internal?
Priority — what matters most? motion quality / content accuracy / brand fidelity / speed?
Variations — one best shot, or 2-3 meaningfully different options (different pacing, energy, or structure — not just color swaps)?

For specific requests ("add a title card", "fix the timing on scene 3"), skip discovery. If the request carries an ongoing autonomous signal ("surprise me", "just build it" — hyperframes-core/references/brief-contract.md § 1), skip discovery too: default to one best shot and state your calls with one-line receipts as you make them.

Step 1 — Design system → `hyperframes-creative`

Establish the visual identity first. If the project has a design spec, read it (precedence frame.md → design.md → DESIGN.md; treat it as brand truth — exact colors, fonts, constraints).

If no spec exists, you MUST read BOTH hyperframes-creative/references/house-style.md AND hyperframes-creative/references/video-composition.md before choosing any color or font. house-style.md gives the "interpret the prompt / generate real content" opener, lazy-default list, and layer recipe; video-composition.md gives the video-medium density / scale / foreground detailing (data bars, registration marks, monospace metadata, "8-10 elements, two the user didn't ask for") that separates "produced" from "generated." Reading only one is the most common miss — video-composition.md is the one agents skip, and it is exactly the one that prevents flat, centered, web-page-looking output. Do not self-invent a palette and skip these; crossing into hyperframes-creative is mandatory here, not an optional branch. From there, also pull a named style/mood → references/visual-styles.md, or the interactive picker → references/design-picker.md, as needed. The spec/style defines the brand, not the composition rules.

Find the angle (vague brief, no spec): before picking colors, write ONE sentence — what does this name/word/topic evoke, and what visual world (metaphor, setting, instrument, motif) expresses it? E.g. a cybersecurity tool → vault doors / perimeter scan lines / lock tumblers; a meditation app → tide, breath, slow light bloom. Read the meaning of the subject, not just its letters; pick a concrete angle over a literal restyle. This is the cheap substitute for prompt expansion (Step 2) on single-scene pieces, where expansion is correctly skipped — and it is the difference between a designed concept and a generic logo-on-a-gradient.

Step 2 — Prompt expansion → `hyperframes-creative`

Run for every multi-scene composition (skip for single-scene pieces and trivial edits). Ground the request against the design spec + house style into a consistent intermediate that downstream work reads the same way. See hyperframes-creative/references/prompt-expansion.md.

Step 3 — Plan

Before writing HTML, think at a high level:

What — the viewer experience: narrative arc, key moments, emotional beats. For a narrated story piece, follow hyperframes-creative/references/story-spine.md — hook in viewer-outcome language, the message landing by beat 2, evidence after.
Structure — how many compositions, sub-comp vs inline, which tracks carry video / audio / overlays / captions. For the monolithic-single-file vs modular-sub-comp call, see hyperframes-core/references/composition-patterns.md § Two Architectures (rule of thumb: ≥3 hard scene cuts, or any reused scene → modularize; a short single-scene piece stays one file).
Rhythm — name the pattern before implementing (e.g. fast-fast-SLOW-SHADER-hold); see hyperframes-creative/references/beat-direction.md.
Timing — which clips drive duration, where transitions land, the pacing.
Layout — build the end state first (see below).
Animate — then add motion via hyperframes-animation.

Layout Before Animation

Position every element where it sits at its most visible moment — fully entered, correctly placed, not yet exiting. Write that as static HTML + CSS first. No GSAP yet.

Why: if you position elements at their animated start state (offscreen, scaled to 0, opacity 0) and tween to where you think they land, you are guessing the final layout — overlaps stay invisible until render. Build the end state first and you see and fix layout problems before adding motion.

Identify the hero frame for each scene — the moment the most elements are simultaneously visible. That is the layout you build.
Write static CSS for that frame. The content container must fill the scene with padding, not absolute offsets:

css

.scene-content {
  display: flex;
  flex-direction: column;
  justify-content: center;
  width: 100%;
  height: 100%;
  padding: 120px 160px; /* padding positions content; fills any scene size */
  gap: 24px;
  box-sizing: border-box;
}

Never use position: absolute; top: Npx on a content container — it overflows when content is taller than the space. Reserve absolute positioning for decoratives.

⚠ The width/height: 100% above only resolves if every ancestor has a resolved height. The root <div data-composition-id> and any wrapper between it and .scene-content must be sized (position: relative; width: 1920px; height: 1080px on the root — see hyperframes-core → "Root must be sized"). Skip this and the flex container collapses to ~0, content piles into the top-left corner, and the first glyph clips at x=0 — while lint/inspect still report 0 issues. And always keep the padding (≥80px) on .scene-content: it is the title-safe margin. Never replace it with bare gap.

Add entrances — animate FROM offscreen/invisible TO the CSS position with gsap.from() (in sub-compositions prefer gsap.fromTo() so the start state is explicit; see hyperframes-core/references/sub-compositions.md). The CSS position is ground truth; the tween is the journey to it.
Exits are transition-handled — per the scene-transition rules in hyperframes-animation/transitions/, only the final scene animates elements out; between scenes the transition IS the exit.

Shared space across time: if element A exits before element B enters in the same area, both still need correct CSS positions for their respective hero frames — timeline ordering keeps them from coexisting, and the layout step catches accidental overlap. Layered glows/shadows and z-stacked depth are intentional overlap; the step is about catching unintentional collisions (two headlines on top of each other, content bleeding off-frame).

Build — delegate to the domain skills

This maps the skill's full surface (see the description) to its references — non-exhaustive; when an intent isn't listed, route through hyperframes-creative (look/concept), hyperframes-animation (motion), hyperframes-core (contract), media-use (audio/captions). The first row is ADDITIVE — read it AND your intent row, not one or the other.

Building…	Read first (in order)
ALWAYS — every non-trivial piece, on top of your intent row below	`hyperframes-creative/references/house-style.md` + `references/video-composition.md` (also gated in Step 1 / HARD-GATE; the "produced, not generated" foreground detailing)
Kinetic typography / text-forward	`hyperframes-animation/techniques.md` (kinetic type) + `adapters/gsap-easing-and-stagger.md` + `rules/kinetic-beat-slam.md`
Title card / lower-third / overlay / PiP / text-behind-subject	`hyperframes-creative/references/composition-patterns.md` + (for the centered/sized frame) `hyperframes-core` → "Root must be sized"
Logo / brand-mark reveal	`hyperframes-animation/rules/svg-path-draw.md` (draw-on) + `rules/3d-text-depth-layers.md` + `rules/scale-swap-transition.md`
Data / stats / numbers	`hyperframes-animation/rules/counting-dynamic-scale.md` + `rules/stat-bars-and-fills.md` + `hyperframes-creative/references/data-in-motion.md`
Product / app / UI demo	`hyperframes-animation/rules/3d-page-scroll.md` + `rules/cursor-click-ripple.md` + `rules/press-release-spring.md`
Audio-reactive / music-driven	`hyperframes-creative/references/audio-reactive.md` (pre-extract bands; map to motion)
Narrated / voiceover / music / SFX / captions	`media-use` → the shared audio engine `scripts/audio.mjs` (one call = TTS + BGM + SFX → `audio_meta.json`); caption authoring + asset placement via `hyperframes-core`. See Audio below.
Multi-scene / transitions	`hyperframes-animation/transitions/overview.md` then `transitions/catalog.md` (you are not done after the overview — the GSAP recipe is in the catalog)
Modular / sub-compositions	`hyperframes-core/references/composition-patterns.md` + `references/sub-compositions.md`

Audio: one engine (TTS · BGM · SFX)

Only when the piece calls for it (per "build exactly what was asked" — no ambient music on a title card). Don't hand-roll TTS or vendor a copy: write a neutral audio_request.json and call the shared engine in media-use. It auto-degrades on one switch — HeyGen credential present → HeyGen TTS + music/SFX retrieval; absent → ElevenLabs/Kokoro TTS, Lyria/MusicGen BGM generation, and the bundled SFX library. Full flag list + request/meta schema: the header comment of media-use/audio/scripts/audio.mjs.

jsonc

// audio_request.json — one line per narrated segment; `id` is yours (joins audio_meta back)
{
  "lines": [
    { "id": "s1", "text": "Your opening line.", "sfx": ["whoosh"] },
    { "id": "s2", "text": "The next beat." },
  ],
  "bgm": { "query": "calm cinematic underscore" }, // omit "mode" → auto (retrieve if HeyGen, else generate); "none" to disable
}

bash

# <MEDIA_DIR> = the installed media-use skill dir (sibling of this skill)
node <MEDIA_DIR>/scripts/audio.mjs --request ./audio_request.json --hyperframes . --out ./audio_meta.json

Then read audio_meta.json: mount each voices[].path + (bgm.path, sfx[]) as <audio> tracks and use voices[].words for captions, all per hyperframes-core (audio tracks + caption authoring). If BGM took the generate path (bgm_pending: true), run media-use/audio/scripts/wait-bgm.mjs before final render.

Output checklist → `hyperframes-cli`

npx hyperframes lint and npx hyperframes validate pass (block on results)
design adherence verified if a spec (frame.md / design.md) exists — checklist in hyperframes-creative/references/design-adherence.md
npx hyperframes inspect passes, or every overflow is intentionally marked
contrast warnings addressed; for multi-scene work, review the animation map (hyperframes-animation/scripts/animation-map.mjs)
deliver the preview; render to MP4 only on explicit request
surface the preview only at handoff (it is the stable, final preview); don't pop one mid-build — build-phase snapshots are headless