html-video Review 2026 — HTML to MP4 for Coding Agents
✅ Pros
- • No per-render fees and no vendor lock-in — Apache-2.0 licensed, runs entirely on your machine with open-source rendering engines
- • 21 curated templates covering data charts, title cards, cinematic intros, kinetic type, and product promos — all single-file HTML, all customizable
- • Works with 14+ coding agent backends — Claude Code, Cursor, Codex, Gemini, Copilot, Grok, Qwen, OpenCode, Aider, and more
- • AI soundtrack generation built in — auto-generate background music and AI narration alongside your visual content
- • Turn an article URL or GitHub repo into a video with one prompt — agent reads the content, storyboards it, and renders a multi-scene video
- • Pluggable rendering engine architecture — Hyperframes shipped as default, Remotion and other engines planned
- • Official project by the Open Design team — well-funded, active community, regular releases
⚠️ Cons
- • Requires ffmpeg and a local coding agent CLI — not a web app, you need the terminal setup
- • Rendering takes place locally, so performance depends on your hardware — complex multi-scene videos can take several minutes on a laptop
- • 21 templates is a good start but still limited compared to professional video template libraries (Canva, Premiere Pro)
- • AI narration quality depends on your TTS setup — no built-in ElevenLabs integration yet, you bring your own voice pipeline
- • Currently macOS/Linux only for full feature set — Windows support for Hyperframes rendering is still experimental
- • Documentation assumes familiarity with coding agents — less accessible for pure video editors who don't use CLI tools
Developers, content creators, and coding agent power users who want to programmatically generate videos from data, articles, or code without paying per-render fees
Free (open-source Apache-2.0) — runs on your machine with no usage charges
html-video Review 2026 — HTML to MP4 for Coding Agents
What Is html-video?
html-video is an open-source tool that turns HTML and CSS into real MP4 video files — entirely on your laptop, with no per-render fees and no vendor lock-in. Created by the Open Design team at nexu-io, it bridges the gap between coding agents and video production: you describe a video (or paste an article link / GitHub repo), and your coding agent generates multi-frame animated HTML that gets rendered to MP4 via Hyperframes + ffmpeg.
Since its launch on May 27, 2026, html-video has gathered 3,332 GitHub stars and 406 forks, riding the wave of programmatic video generation for AI-assisted content creation. It ships with 21 curated templates, pluggable rendering engines, and optional AI soundtrack generation — all under the Apache-2.0 license.
Why This Exists
Programmatic video generation (HTML→video) is a real category, but every existing engine has been opinionated, proprietary, or both:
| Engine | Cost | License | Paradigm |
|---|---|---|---|
| Hyperframes (default engine) | Free | Apache-2.0 | HTML + CSS + GSAP, agent-skill driven |
| Remotion | Free up to 4 devs | Source-available | React components |
| Motion Canvas | Free | MIT | TypeScript + Canvas API |
| Descript | $24/mo+ | Proprietary | Timeline-based editor |
| Canva | Free tier / $13/mo | Proprietary | Drag-and-drop design |
html-video’s insight is simple: let the coding agent author the HTML, and let open-source engines render it. The agent handles the creative work — storyboarding, writing CSS animations, composing scenes — and html-video handles the mechanical work of frame capture and encoding.
Core Architecture
html-video operates in two phases:
-
Authoring Phase: Your coding agent (Claude Code, Cursor, Codex, etc.) generates a multi-scene HTML file using one of the 21 included templates or a custom design. Each scene is a self-contained HTML frame with CSS animations.
-
Rendering Phase: The Hyperframes engine (included, shipped) loads each HTML scene in a headless Chromium browser, captures frames at the configured frame rate, and stitches them into an MP4 via ffmpeg. The output is a real, playable video file.
This split means the tool is render-engine agnostic — Remotion, Motion Canvas, and other engines are planned as pluggable backends. You’re not locked into any single authoring model.
Template Library: 21 Ready-to-Use Scenes
The template library covers the most common video scene types:
| Template Category | Examples |
|---|---|
| Data Visualization | frame-data-chart-nyt (NYT-style animated line chart), Swiss-grid data cards, Vignelli-style data cards |
| Title Cards | frame-glitch-title (chromatic aberration + scanlines), frame-liquid-bg-hero (aurora gradient), kinetic type, neon titles |
| Cinematic | frame-light-leak-cinema (warm film grain + light leak), warm-grain editorial |
| Product & Promo | Multi-scene product promo, Takram-organic motion, decision-tree explainers |
| Outros & Effects | frame-logo-outro (animated logo end card), vfx-text-cursor (typewriter + blinking cursor), VFX transitions |
Each template is a single-file HTML with embedded CSS animations — no build step, no external dependencies. Drop one into an agent prompt, let it fill in your content, and export to MP4.
Coding Agent Support
html-video supports 14+ coding agent backends:
Fully supported: Open Design CLI, Windsurf CLI, Trae CLI, Claude Code, Cursor, Codex CLI, Gemini Code Assist, Grok CLI, Qwen Coder, OpenCode, Copilot CLI, Aider, Hermes, Anthropic API
The agent workflow looks like this:
$ claude "Create a product launch video for my new app. Use the multi-scene product promo template."
→ Agent writes HTML scenes with your content
→ Agent calls html-video renderer
→ MP4 lands in your project directoryAI Soundtrack & Narration
html-video can add audio to your videos through pluggable pipelines:
- Background music: Auto-generated AI music tracks via MiniMax or other providers
- Narration: AI text-to-speech voiceover that syncs with scene timing
- Sound effects: Scene-triggered audio cues for transitions and emphasis
The audio layer is optional and provider-dependent — you configure which TTS/music services to use, and the agent weaves them into the final render.
Link-to-Video: One Prompt Magic
One of html-video’s most impressive features is turning a URL into a video:
$ claude "Turn https://example.com/blog/article into a 30-second video"The agent:
- Fetches and reads the article content
- Storyboards it into 3-5 scenes
- Applies appropriate templates (title card, data highlights, quotes, outro)
- Renders the full MP4
This works with article links, GitHub repos, and even raw data — making it a powerful tool for automated content repurposing.
Real-World Use Cases
Developer documentation videos: A team maintains 50 microservices, each with a README. They write a script that watches for README changes and auto-generates a 20-second “what’s new” video using html-video’s link-to-video feature. Every PR update produces a video summary.
Data journalism: A data analyst generates daily animated charts from their pipeline output. The NYT-style data chart template + AI narration produces shareable video summaries without touching a video editor.
Product launch videos: An indie hacker describes their new SaaS product to Claude Code, which generates a multi-scene promo video using the liquid-bg-hero template for the opener, feature cards for the body, and the logo outro. Total time: 15 minutes instead of 2 days.
AI-generated YouTube shorts: A content creator writes an agent that reads HN front-page articles, summarizes the top story, and produces a 60-second short with kinetic type and AI narration — posted daily without manual editing.
How It Compares
| Feature | html-video | Descript | Canva | Remotion |
|---|---|---|---|---|
| License | Apache-2.0 | Proprietary | Proprietary | Source-available |
| Per-Render Cost | $0 | $24-40/mo | $13-30/mo | Free (≤4 devs) |
| Agent-Powered | ✅ Native | ❌ | ❌ | ❌ (React only) |
| Templates | 21 curated | Hundreds | Thousands | Community |
| Hardware | Your laptop | Cloud + local | Cloud | Your laptop |
| AI Narration | ✅ Pluggable | ✅ Built-in | ✅ Built-in | ❌ |
| URL→Video | ✅ | ❌ | ❌ | ❌ |
Pricing
html-video is completely free under the Apache-2.0 license. There are no per-render fees, no subscription tiers, and no usage limits. The only costs are:
- Your coding agent subscription (Claude Code, Cursor, etc. — typically $10-20/mo)
- Local compute (rendering uses your CPU/GPU)
- Optional TTS/audio service API fees if you use third-party providers
Verdict
html-video fills a genuine gap in the AI tools ecosystem: programmatic video generation that respects your wallet and your workflow. By letting coding agents author the creative content and open-source engines handle the rendering, it democratizes video production in the same way that static site generators democratized web publishing.
The 21 templates are polished and genuinely useful, the agent integration is smooth, and the Apache-2.0 license means you can build a business on top of it without asking for permission. The main limitations are local performance (rendering is CPU-bound) and the requirement for CLI comfort — this is not a tool for Canva-refugees who want drag-and-drop.
For developers, content operators, and AI workstation users, html-video is a powerful addition to the toolkit. It turns a one-line prompt into a shareable video, and that kind of leverage is exactly what AI tools should provide.
Rating: 8.0/10 — Polished open-source tool that elegantly solves a real problem. Missing a web UI for non-CLI users, but that’s a deliberate design choice, not a bug.