Guide · Agent skill
How to use the quietshrink Video Compressor agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The quietshrink skill teaches an AI agent to compress large macOS screen recordings (.mov or .mp4) on Apple Silicon Macs using the built-in hardware HEVC encoder, so the files are easier to share in chat, email or GitHub. This guide explains what an agent skill is, how to install quietshrink, how to ask for compression in plain language, and what to check before running it. By the end you will be able to install the skill, pick a quality preset and know when the skill is the wrong tool.
quietshrink Video CompressorHKUDS/CLI-Anything · Apache-2.0 Download the skill Quick start
- 1Check your Mac. Confirm you have macOS on Apple Silicon (M1 or later) for the best results.
- 2Install ffmpeg. Run brew install ffmpeg so the hardware HEVC encoder path is available.
- 3Install the skill. Download the quietshrink ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/, or upload it under Settings → Capabilities → Skills.
- 4Verify the environment. Run cli-anything-quietshrink doctor to confirm the CLI and ffmpeg are working.
- 5Ask for compression. Tell the agent which file to shrink and what it is for, such as sharing in chat or keeping it for docs.
- 6Check the result. Review the input and output sizes and the preset reported, then keep the original until you confirm the copy looks right.
What the quietshrink skill makes the agent do
An agent skill is a folder containing a SKILL.md file that tells an AI agent when and how to use a tool. The quietshrink skill is the instruction set for cli-anything-quietshrink, a command-line tool for compressing video files on macOS Apple Silicon.
The skill uses the Mac's Media Engine (the hardware HEVC encoder) rather than the CPU, which the source says makes encoding fast and silent. The description claims typical file size reductions of 70-90% while staying visually lossless; the skill's own preset table lists typical reductions of about 84-90% depending on the preset.
- Compresses screen recordings and other .mov or .mp4 files
- Offers four quality presets, from smallest file to near-source quality
- Inspects a video's codec, resolution, framerate and duration before compressing
- Verifies the environment before processing
Install the quietshrink skill
You can install the skill in one of three places. Download the quietshrink ZIP from this page and unzip it into one of the following locations, depending on how you use your agent.
For Claude Code, put the unzipped skill folder in ~/.claude/skills/ to make it available everywhere, or in a project's .claude/skills/ folder to limit it to that project. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal (all projects): ~/.claude/skills/
- Single project: .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Install the CLI and its dependencies
The skill calls the cli-anything-quietshrink command, so that tool and ffmpeg must be present. The source says ffmpeg must include hevc_videotoolbox support and is installed via Homebrew. Run doctor after installing to check the environment.
Requirements: macOS on Apple Silicon (M1 or later) gives the best results. Other systems fall back to a less efficient software path.
- Install ffmpeg with Homebrew: brew install ffmpeg
- Check the setup: cli-anything-quietshrink doctor
brew install ffmpeg
cli-anything-quietshrink doctorTrigger the skill with natural requests
You do not need to name commands. The skill is meant to activate when you ask to compress or shrink a screen recording, make a file small enough to share, or reduce a video for chat, email or GitHub. Name the file and say what it is for, because the preset depends on that.
If you say the recording is important or for documentation, the agent should use the default transparent preset. If you need the smallest file for quick sharing, ask for tiny. If you need the file for archive or editing, ask for pristine.
- "Shrink ~/Desktop/demo.mov so I can email it."
- "Compress this screencast for the docs, keep it visually lossless."
- "Make a tiny version of recording.mp4 for Slack."
- "What codec and resolution is this screen recording?"
Understand the quality presets
Each preset trades file size against quality. The table below comes from the skill's documentation. Typical reduction and SSIM figures are the skill's own estimates, not independent tests.
The default, transparent, is the one to use for anything important. Pristine is still lossy, so keep your original if you must retain the exact source.
- tiny (q 50): about 90% reduction, SSIM about 0.95, for chat and email where small artifacts are acceptable
- balanced (q 55): about 88% reduction, SSIM about 0.99, for docs and sharing
- transparent (q 60, default): about 87% reduction, SSIM about 0.99+, for important material
- pristine (q 70): about 84% reduction, SSIM about 0.997, for archival near-source quality
cli-anything-quietshrink presets
cli-anything-quietshrink compress -q tiny recording.mov
cli-anything-quietshrink compress -q transparent recording.mov
cli-anything-quietshrink probe recording.movRead the output and handle errors
The compress command reports input and output size, saved bytes and percent, duration, elapsed time, encoding speed and the preset used. All commands accept --json for machine-readable output, which is useful when an agent or script needs to parse results.
If something fails, the skill lists three known errors with fixes. For compression_failed, check that the input is not corrupted and try verbose mode to see ffmpeg errors.
- ffmpeg not found: run brew install ffmpeg
- hevc_videotoolbox not available: run brew reinstall ffmpeg
- compression_failed: check the input file and retry with --verbose
brew reinstall ffmpeg
cli-anything-quietshrink compress --verbose recording.movSafety notes: scripts and network
The skill runs scripts, because it calls the cli-anything-quietshrink command and its install script. Review the scripts in the ZIP before installing if you need to know exactly what they do. The listed network use is none, and the skill's files are SKILL.md only as listed on this page.
Compression is lossy. Keep your original recording until you have checked the compressed copy, especially for archival material.
- Runs scripts: yes (the CLI and its install script)
- Uses network: no, according to this page's assessment
- Keep originals for archival or editing work
What people use it for
- Shrinking a long screencast so it fits an email attachment limit
- Reducing a bug-report recording before posting it to a GitHub issue
- Making a quick tiny copy of a demo for chat
- Inspecting a recording's codec, resolution and framerate before deciding how to compress it
Frequently asked questions
Is the quietshrink skill free?
This page does not list pricing or licence terms for the skill itself. The listed licence is Apache-2.0 for the HKUDS/CLI-Anything repository that contains it, so check that repository's licence file for the terms that apply.
What do I need to run quietshrink?
You need macOS on Apple Silicon (M1 or later) for best results, ffmpeg with hevc_videotoolbox support installed via Homebrew, and the cli-anything-quietshrink CLI with its install script. Other systems use a less efficient software path.
Does quietshrink work on Intel Macs or Windows?
The skill says non-Apple Silicon systems fall back to a software path that is less efficient, and the hardware encoder is required for best results. Expect slower encoding and smaller savings there.
Should I use quietshrink for camera footage or vlogs?
The skill's own documentation says not to use it on non-screen content, because savings are much smaller without duplicate frames to drop. Use it mainly for screen recordings.
Is the compression lossless?
No. Compression is lossy, even at the pristine preset, so keep your original file if you need the exact source.
Why does the agent say hevc_videotoolbox is not available?
Your ffmpeg build lacks the hardware HEVC encoder. The skill's fix is brew reinstall ffmpeg, then run cli-anything-quietshrink doctor again.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
quietshrink suits Mac users on Apple Silicon who record screencasts and need smaller files for chat, email or GitHub without fan noise. Skip it for camera footage, for non-Apple Silicon machines where you need efficiency, or for archival originals that must stay untouched.
Download the skill