# quietshrink Agent Harness Agent-native CLI for [quietshrink](https://github.com/achiya-automation/quietshrink) — compress macOS screen recordings on Apple Silicon with zero CPU stress, achieving 70-90% file size reduction at visually lossless quality. ## What this harness does Wraps the standalone `quietshrink` bash CLI with a Python (click-based) interface that exposes structured JSON output for AI agents. Agents can: - **`compress`** a video file with quality presets (tiny / balanced / transparent / pristine) - **`probe`** a file before compressing to inspect codec, resolution, duration - **`presets`** list available quality profiles with empirical SSIM/size data - **`doctor`** verify that ffmpeg, hevc_videotoolbox, and the bash CLI are ready All commands accept `--json` for machine-readable output and exit with proper error codes. ## Why agents care Screen recording compression is a frequent agent task: "compress this screencast before sharing", "make this file smaller", "convert recording.mov for email". The agent needs deterministic, predictable behavior: - Hardware encoding → stays under any budget; computer remains responsive - Smart frame deduplication → exploits the static nature of screen content - Long GOP + adaptive quantization → matches software-encoder size at hardware speed - SSIM-validated quality presets → the agent can pick a preset based on the user's goal (sharing vs archiving) ## Install ```bash pip install git+https://github.com/HKUDS/CLI-Anything.git#subdirectory=quietshrink/agent-harness ``` The harness depends on the bash `quietshrink` CLI being available in `$PATH`. Install it with: ```bash curl -fsSL https://raw.githubusercontent.com/achiya-automation/quietshrink/main/install.sh | bash ``` ## Usage ```bash # Inspect a file cli-anything-quietshrink probe recording.mov --json # Compress (default: transparent quality) cli-anything-quietshrink compress recording.mov compressed.mov --json # List quality presets cli-anything-quietshrink presets --json # Verify environment cli-anything-quietshrink doctor --json ``` ## Source - Main repo: - Why this approach: - License: MIT