Tooling Review

HyperFrames — write HTML, render video

An open-source engine (HeyGen, Apache 2.0) that turns an HTML file with timing attributes into a deterministic MP4. Installed on Malik as 20 agent skills. This page is a map of everything it can do — with live samples.

One line: you write a webpage, it turns into a video.
20 skills installed 1 router · 11 workflows · 8 capability layers MP4 or transparent overlay 16:9 · 9:16 · any length 6 samples rendered from our own content ✓

Our own samples · rendered on Malik just now

Every clip below was authored as an HTML file and rendered to video on our server — using our own content and data, not HeyGen's demos. Most are deliberately unbranded / neutral to prove the flexibility; one shows the Accord brand option. Hover to play.

All six above were made from our material — Burn Watch runway, the SAFE ledger, our brand chevron — authored as HTML and rendered locally. Total build: minutes. Same tech I already use for decks and dashboards, now with a time axis.

HeyGen reference samples

For range: HeyGen's own showcase pieces showing what a fully-produced HyperFrames video looks like (3D, shader transitions, footage compositing, voiceover). Each has open source you can inspect and remix.

All 20 skills

One router reads your request and dispatches to a workflow. Workflows compose against the atomic capability layers. You never manage this by hand — you just say what you want.

1

The router

read-first
/hyperframes
Reads any "make / edit / animate / render a video" request and routes it to the right workflow below. The entry point — invoked automatically.
2

Creation workflows

11 · "make me a…"
Each is a full production pipeline: plan → source assets → design → build HTML → render.
/product-launch-video
Launch / promote a product — from a URL, brief, or script. Up to ~3 min (sweet spot 30–90s).
/website-to-video
Turn a general website into a video — site tour, portfolio showcase, social clip from the site's own visuals.
/faceless-explainer
Explain a topic from text — no product or site; every visual is invented (typography, diagrams, data-viz).
/pr-to-video
A GitHub PR → changelog / feature-reveal / fix explainer with animated code diffs.aljeel repo wired — usable now
/embedded-captions
Add captions / subtitles to an existing talking-head video — footage untouched.
/talking-head-recut
Package an interview / podcast with designed overlays — lower-thirds, data callouts, kinetic titles, pull-quotes, PiP.
/motion-graphics
Short, unnarrated, design-led piece (<10s) — kinetic type, a stat hit, a logo sting, a chart hit.used for our test render
/music-to-video
A music track → beat-synced video — lyric, slideshow, or kinetic promo; music drives pacing.
/slideshow
A pitch deck / interactive deck — discrete slides, fragments, branching, hotspots, presenter mode. Output is a navigable deck, not MP4.
/general-video
Anything else — longer or multi-scene pieces, brand / sizzle reel, title card, static loop, freeform.
/remotion-to-hyperframes
Port an existing Remotion (React) composition to HyperFrames HTML. One-way migration.
3

Capability layers

8 · atomic powers
The building blocks the workflows compose against — pulled directly when a specific layer is needed.
/hyperframes-core
The composition contract — timing attributes, clips, tracks, sub-compositions, variables, determinism rules.
/hyperframes-animation
All motion — GSAP, Lottie, Three.js (3D), Anime.js, CSS, WAAPI, WebGPU; scene blueprints + transitions.
/hyperframes-keyframes
Frame-accurate keyframes — SVG path-draw / morph, masks, FLIP, 3D depth, seek-safe motion across runtimes.
/hyperframes-creative
Creative direction — frame.md / design.md, palettes, typography, narration, beat planning, audio-reactive visuals.
/hyperframes-media
Audio + media — TTS voiceover, background music, SFX, Whisper transcription, background removal, captions.
/media-use
Resolve any media need (BGM / SFX / image / icon) into a frozen local file with manifest tracking.
/hyperframes-cli
Dev loop — init, lint, validate, preview, render, publish, doctor + AWS Lambda distributed rendering for big jobs.
/hyperframes-registry
Install pre-built blocks — animated charts, chart-races, map animations, shader transitions, social overlays.

What we could actually make

Branded or unbranded — your call, per project. A sampling mapped to things we already do.

📊

Data-viz / chart-race

Burn Watch, cash simulator, benchmark data → a chart that draws and animates itself.

📈

Investor / LP video

LP update or deck as a 60–90s piece, paired with our ElevenLabs voiceover.

🔧

PR → changelog video

Feature-reveal videos straight from a GitHub PR — aljeel repo or any code work.

🚀

Product explainer

Regent / Mushir / Accord OS reveals — UI mockups, narration, feature beats.

📱

Social clips

9:16 vertical for Reels / Shorts, or 16:9 — kinetic captions, overlays, music.

🎞️

Transparent overlays

.mov / .webm lower-thirds and logo stings to drop onto other footage.

🖼️

Interactive decks

Branching + hotspots + presenter mode — a navigable deck, no render needed.

✍️

Kinetic type / headlines

Animated tweets, headlines, quotes — quick motion hits, seconds to render.

The flexibility knobs

Nothing is locked. Branding is opt-in per project — not baked into the engine.

🎨 Branding — opt-in, per project

A frame.md file is what applies a brand. Drop it in → Accord look (Newsreader serif, navy, gold). Leave it out → a neutral blank canvas styled freely per project. Or write a different frame.md for Regent, Mushir, a client.

Accord frame.mdno frame (neutral)Regent frame.mdclient frame.md

📐 Format — set per render

Aspect ratio, length and output type are chosen at build time. Same composition can ship 16:9 for a deck and 9:16 for social, or export as a transparent overlay instead of a solid MP4.

16:99:1610s → 3 minMP4transparent .mov / .webm
Recommendation: keep branding off by default — I only apply an Accord (or Regent / client) frame.md when you say so. Everything else stays a free canvas.