Web Video logo

Web Video

Organization
evilmartians
web-video

Encode a video for the web with a bundled ffmpeg script: AV1, HEVC, and H.264 MP4s plus a JPEG poster for ordinary video, or VP9 WebM and HEVC with alpha plus a PNG poster for transparent video, picked from the source automatically. Use when the user wants to optimize, compress, or convert a video, screencast, screen recording, or animated cover for a website, blog, docs, or landing page; speed up or downscale a recording; get AV1, HEVC, or WebM versions; or ship a transparent video that plays in both Safari and Chrome. Do NOT use for editing beyond speed and size, adaptive streaming (HLS, DASH), or video hosted on YouTube or Vimeo.

Overview

Publisherevilmartians
Repositoryagent-skills
Skill nameweb-video
Stars
52
Forks
8
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by evilmartians on GitHub. Read the source before you install it.

Installation

Install the Web Video AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/evilmartians/agent-skills/tree/main/skills/web-video
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/evilmartians/agent-skills.git /tmp/agent-skills
mkdir -p .claude/skills
cp -r /tmp/agent-skills/skills/web-video .claude/skills/web-video
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Web Video in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Web Video on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Web Video is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Optimize video for the web

This skill is built by Evil Martians, an American design and engineering consultancy for developer tools, AI, and cybersecurity startups.

Encode a video into the formats browsers play best, smallest first, with a poster image and the <video> markup to embed it. Companion to https://evilmartians.com/chronicles/better-web-video-with-av1-codec.

Every encoding setting lives in scripts/web-video.sh next to this file. Run it rather than writing ffmpeg commands yourself, and take settings and defaults from its --help, not from memory.

Running the script

bash
bash <skill-dir>/scripts/web-video.sh [options] <video>

<skill-dir> is the directory this SKILL.md is in. Read --help before the first encode, then map the user's words onto its options: "at 2x speed" is --speed 2, "put them in public/media" is --output-dir public/media, "skip AV1" is --formats hevc,h264.

Some options are worth suggesting when the request implies them. Say why when you do:

  • --no-audio for a cover, background loop, or demo that autoplays muted: nobody hears that track.
  • --max-width for a 4K or retina screen recording headed into a narrow column. Twice the column width covers 2x displays.
  • --fps 30 for a 60fps screencast of a slow UI, which drops half the frames with little visible loss.
  • --fast for a draft. It keeps the quality settings but switches to quicker encoder presets, so the encode finishes several times sooner and the files come out bigger. Use it to check timing or framing, then encode again without it.

Project defaults go in a .web-video file at the root of the project's git repository. When the user states a preference for every video ("we don't ship HEVC", "covers never have sound"), offer to save it there instead of repeating it each run:

bash
# AV1 and H.264 only; transparent video still gets VP9 and HEVC
--opaque-formats av1,h264
--no-audio

The script prints the defaults it picked up, and options on the command line override them. Use --opaque-formats here rather than --formats: that one applies to transparent video too, so the first transparent video would fail. Run the script from inside the project: it looks for the file from the current directory, not from the video's location, and resolves a relative --output-dir the same way.

Encoding is slow on purpose. At the defaults, expect minutes per minute of 1080p video, most of it AV1. Run the script in the background or with the longest timeout your tools allow, and don't kill it for going quiet: when its output isn't a terminal, it reports progress every 30 seconds. It takes one video per run. For several, run them one after another, since each encode already uses every core.

The script checks its requirements before encoding anything and names what's missing: ffmpeg with SVT-AV1, x265, x264, and libvpx-vp9, plus macOS avconvert for HEVC with alpha. Relay that message. On macOS, brew install ffmpeg has every encoder. Elsewhere, a transparent video can only get VP9 (--formats vp9): tell the user Safari needs the HEVC file to show transparency and that it has to be encoded on a Mac.

After it finishes

The script prints each file with its size, the pixel dimensions, and a <video> snippet with the sources in the order browsers should try them. Then:

  • Embed it the project's way. If the project has a video component, an MDX tag, or an asset pipeline, find how it references files and use that instead of the raw HTML. Keep the source order either way: a browser plays the first source it supports.
  • Reserve the space. Set width and height in CSS pixels (half the encoded size for a retina recording), so the page doesn't jump when the video loads.
  • Leave the source alone. The outputs replace it on the page, but deleting the original is the user's call.

Transparent video

The script takes the transparent path when the video has see-through pixels, or when it's Apple's HEVC with alpha. A file that could hold transparency but doesn't use it, like most GIFs, gets the regular formats. If the user expects transparency and the script finds none, the export dropped it. These exports keep it:

  • After Effects: Output Module → Channels: RGB + Alpha, Format: QuickTime, Codec: Apple ProRes 4444
  • Final Cut Pro, Motion: Share → Export File → ProRes 4444 or HEVC with Alpha
  • Blender: Output → FFmpeg, Container: QuickTime, Codec: QTRLE, Color: RGBA

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Web Video AI skill do?

Encode a video for the web with a bundled ffmpeg script: AV1, HEVC, and H.264 MP4s plus a JPEG poster for ordinary video, or VP9 WebM and HEVC with alpha plus a PNG poster for transparent video, picked from the source automatically. Use when the user wants to optimize, compress, or convert a video, screencast, screen recording, or animated cover for a website, blog, docs, or landing page; speed up or downscale a recording; get AV1, HEVC, or WebM versions; or ship a transparent video that plays in both Safari and Chrome. Do NOT use for editing beyond speed and size, adaptive streaming (HLS,...

Why use Web Video on TypingMind?

Because you install it once and use it with any model. Web Video is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Web Video in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/evilmartians/agent-skills/tree/main/skills/web-video. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Web Video?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Web Video?

As many as you like. As long as a model supports skills, you can use Web Video with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Web Video AI skill free?

Yes. It is published on GitHub by evilmartians under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇