html-video
A CLI for programmatic video from HTML: the agent assembles frames from templates and renders a finished MP4 on your machine
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Runs Chromium via Playwright and a local render engine
- Passes frame content to the selected agent, which can execute commands on the machine
Install
Manual install
git clone https://github.com/nexu-io/html-video
cd html-video
pnpm install
pnpm -r buildThis is third-party code. Review the repository files before installing.
What it does
html-video runs a local studio where an agent assembles a video from HTML and CSS frame by frame, using one of 21 ready templates or starting from scratch. The source can be a text description, an article link, or a GitHub repository, and rendering happens locally through pluggable engines without paying a third-party rendering service. You can add a soundtrack: MiniMax generates background music or narration from an API key. The tool detects agents already installed on the system, such as Claude Code, Cursor, or Codex CLI, and drives them through a single agent loop.
Who it is for. For developers and content teams who need to turn text, articles, or data into short videos without a video editor.
Good fit when
- You need to quickly assemble a demo or product announcement from a ready template
- You need to turn an article or a GitHub repository into a narrated video
- You need to render video locally without a cloud service subscription
Not a fit when
- You need a full video editor with frame-by-frame manual editing
- You have no supported agent or Anthropic API key on the machine
Example request
Assemble a video from this article using the frame-data-chart-nyt template and add background musicLimitations
Rendering needs Node.js, pnpm, and Chromium for Playwright, and everything is built locally from source. Narration and music require a separate paid MiniMax key; without it you only get the silent version of the video. The fallback scenario without any installed agent goes through the Anthropic API, which typically needs a VPN from Russia.
How to disable. Stop the studio and delete the html-video repository folder; the install does not touch anything outside its own directory.
Security check
- Runs Chromium via Playwright and a local render engine
- Passes frame content to the selected agent, which can execute commands on the machine
README in short
The README presents html-video as a tool for programmatic video creation right on a developer's machine. An agent assembles a clip from HTML frames using one of 21 templates, with a prompt, article, or repository as the source. The project works through pluggable render engines, supports 14 different agent backends, and offers optional narration via MiniMax. It is Apache-2.0 licensed, an official project from the Open Design team.
FAQ
Do I need a subscription to render videos?
No, rendering happens locally through pluggable engines; only narration and music via MiniMax are optionally paid.
Which agent does it use by default?
The studio leads with Open Design (Vela) and falls back to the first available agent already installed on the machine.
Related
Skills and CLI from HeyGen: the agent writes HTML compositions and renders them into MP4 videos, captions, motion graphics and decks
Blender MCP
MCP for Blender
An MCP server and Blender addon: the agent creates and edits objects, materials and scenes, pulls assets and runs Python in Blender
Official Remotion skills for making videos in React: compositions, animation, captions, maps and rendering
Agentic video studio for Claude Code, Cursor and Codex: script, voiceover, images, stock footage, music, subtitles and render