Video, addressed to something that types.

Every video tool before this one had a timeline you drag. This description names a format an agent already writes, then names the output, then names the reader. Nothing in the six words describes a feature.

$ npm i hyperframes

github.com/heygen-com/hyperframes

Write HTML.The input format. An agent can already produce it, so no new syntax has to be learned by either party.

the input

Render video.The output. A file, not a preview, and a format the tool has no control over once it leaves.

the output

Built for agents.The reader. This is the only sentence that makes a choice rather than stating a fact.

the audience

what changes if you drop sentence threeit becomes an HTML to video renderer which already exists

HTML is the interesting choice

A video DSL would need documentation and a parser. HTML arrives with a layout engine, a text renderer, and a model that agents have written more of than any other format.

Three full stops in six words

Each sentence is one clause with a period after it. That is a deliberate rhythm, and it reads as a spec line rather than a pitch.

What is not on this page

No codec, no resolution, no render time, no supported subset of HTML. The trending description names none of it.

Concept mock, not affiliated with the project. The complete GitHub Trending description on 8 September 2026 is "Write HTML. Render video. Built for agents." The three rows are those three sentences in order, labelled by what each one names. No format support, performance figure, or output sample is represented here.