General Video

media-use: Before sourcing audio/images, call /media-use to resolve BGM/SFX/images from the HeyGen catalog. 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.

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.

<HARD-GATE>
Before writing ANY composition HTML, verify you have ALL FOUR:
1. **A visual identity** grounded in the spec or `house-style.md` — not invented on the spot. (Reaching for `#333`, `#3b82f6`, or `Roboto`? You skipped it.)
2. **A one-sentence concept angle** (the "find the angle" step) for anything beyond a trivial edit — not a literal restyle of the prompt words.
3. **A font pairing from the embed list** (`hyperframes-creative/references/typography.md` → "Fonts that embed") chosen on purpose — not `Inter`/`Helvetica Neue`/`system-ui` by default, and never an un-embedded display font you're just hoping renders (un-bundled names embed only if auto-captured locally — and cloud renders won't capture them).
4. **A foreground/density plan from `video-composition.md`** — the anchor-to-edges, 8-10-elements, foreground-metadata, background-texture rules. (Centered stack on a flat color with fewer than ~6 elements and no edge-anchored detail? You skipped it — that is the generic tell.)
</HARD-GATE>

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:

1. What — the viewer experience: narrative arc, key moments, emotional beats.
2. 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).
3. Rhythm — name the pattern before implementing (e.g. fast-fast-SLOW-SHADER-hold); see hyperframes-creative/references/beat-direction.md.
4. Timing — which clips drive duration, where transitions land, the pacing.
5. Layout — build the end state first (see below).
6. Animate — then add motion via hyperframes-animation.