HyperFrames is HeyGen's open HTML-to-MP4 engine plus a loadable skill pack. Docs live at hyperframes.heygen.com/introduction. The repo heygen-com/hyperframes is Apache-2.0, created 2026-03-10, TypeScript. On 2026-09-10 GitHub listed 48,513 stars and 4,443 forks, last push the same day. README: write HTML, render video, built for agents. Playground: hyperframes.dev.
Compare Generate Image if you only needed stills, Slack GIF Creator if you needed a short GIF, or Claude Code if you needed the host that loads /hyperframes.
Key Features
- HTML compositions: timing via
data-start,data-duration, tracks, and seekable animation adapters (GSAP, CSS, Lottie, Three.js, Anime.js, WAAPI). No React requirement, no bundler for a plainindex.html. - 20 published skills: README says read
/hyperframesfirst. It routes product-launch, faceless-explainer, pr-to-video, motion-graphics, slideshow, remotion-to-hyperframes, and more. Core set installs on demand;--allpulls every published skill. - CLI loop:
npx hyperframes init,preview,lint,render. Cloud render on HeyGen and AWS Lambda paths are documented. Requires Node.js 22+ and FFmpeg. - Deterministic capture: headless Chrome seeks each frame, FFmpeg encodes. Same input, same MP4.
- frame.md: a DESIGN.md superset so an agent can compose from brand tokens without guessing web chrome.
Limitation: 48,513 stars is a heat signal, not an audit. README warns npx skills add can lag main by hours; use npx hyperframes skills update for the current core set. FFmpeg and Node 22 are hard requirements. Apache-2.0 covers the engine; HeyGen cloud render still bills HeyGen if you use that path.
Use Cases
- Claude Code / Codex / Cursor users who want a product intro or PR walkthrough as MP4 without opening After Effects.
- Teams with a design.md who need the same tokens on a 1920x1080 frame.
- People who should use Generate Image instead if they only need a still.
Pricing
| Piece | Price | Notes from first-party pages 2026-09-10 |
|---|---|---|
| hyperframes engine and skills | $0 | Apache-2.0. About 48,513 stars. README: no per-render fees, no commercial-use threshold. |
| Host agent | Confirm that product | Claude Code, Codex, Cursor, Gemini CLI. |
| HeyGen cloud / Lambda | Confirm HeyGen or AWS | Optional. Local render is the default. |
Getting Started
npx skills add heygen-com/hyperframesand pick the Core Skills group, ornpx hyperframes skills updatefor a non-interactive core install.npx hyperframes init my-video && cd my-video && npx hyperframes preview.- Ask the agent: using
/hyperframes, create a 10-second product intro with a fade-in title. npx hyperframes renderwhen the preview is right.
First-party resource: Quickstart and the README.
Frequently Asked Questions
Is this HeyGen the SaaS?
No. This is the Apache-2.0 renderer and skill pack. HeyGen the product can host cloud renders. Local render does not need a HeyGen account.
Do I have to install all 20 skills?
No. README: the picker starts with nothing selected. Core Skills is enough. --all is deliberate.
How is it different from Remotion?
README comparison: Remotion authors in React; HyperFrames authors in HTML. Remotion License is source-available; HyperFrames is Apache-2.0.
Alternatives
- Generate Image: still image generation skill.
- Slack GIF Creator: short GIF, not MP4 pipelines.
- Claude Code: the coding agent that can load this pack.
Tips
- Quote Apache-2.0, 48,513 stars, Node 22+, FFmpeg, and 20 skills from pages checked 2026-09-10.
- Prefer
npx hyperframes skills updateover a stale skills.sh blob. - Keep animations seekable. Wall-clock-only GSAP will drift in the renderer.
Conclusion
HyperFrames is the HTML-native video stack agents can actually write: twenty skills, a CLI, and Apache-2.0 local renders. Start at the docs. If you only needed a still, open Generate Image instead.
Comments
No comments yet. Be the first to comment!
Related Tools
Related Insights
Seven AI Coding CLIs, Six Months: No Matter How Strong the Model, Work Needs Supervision
Claude Code, Codex, opencode, pi, omp and DeepSeek Harness all have personalities. After six months of deep use I run a division of labor: pi for the fastest cheapest reviews, omp for complex PRs, DeepSeek Harness on V4 Flash for high-frequency low-cost review, and Claude Code, Qoder and Cursor for writing. No matter how strong the model, work needs supervision — ideally from an independent third party.

Anthropic Subagent: The Multi-Agent Architecture Revolution
Deep dive into Anthropic multi-agent architecture design. Learn how Subagents break through context window limitations, achieve 90% performance improvements, and real-world applications in Claude Code.
Skills + Hooks + Plugins: How Anthropic Redefined AI Coding Tool Extensibility
An in-depth analysis of Claude Code's trinity architecture of Skills, Hooks, and Plugins. Explore why this design is more advanced than GitHub Copilot and Cursor, and how it redefines AI coding tool extensibility through open standards.