How do I install the Ps Image Compression skill?

Run npx skillmds@latest add psmyrdek/ps-image-compression in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Ps Image Compression safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Ps Image Compression?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Ps Image Compression free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Ps Image Compression?

psmyrdek (@psmyrdek) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Ps Image Compression by psmyrdek — Agent Skill

Name: Ps Image Compression
Rating: 4.5 (4 reviews)
Author: psmyrdek

Ps Image Compression

Compress PNG images with pngquant, optionally downscaling first, driven by a quality or resolution parameter. Use when the user says "compress this image", "make this png smaller", "optimize screenshots", "resize to 1200px", or invokes /ps-image-compression.

psmyrdek Updated Sep 22, 2026

File contents

Compress PNG Images

Shrink PNG files using pngquant (lossy palette quantization), with optional downscaling via sips before quantization. Requires pngquant (brew install pngquant); sips ships with macOS.

Inputs

Target: one or more PNG files, or a directory/glob. If the user didn't name one (or it doesn't exist), list PNGs in the current directory and ask them to pick.
quality (optional): target visual quality, given as a number 0-100 or a range like 65-85. Maps to pngquant's --quality=min-max.
resolution (optional): max dimension in pixels (e.g. 1200) or an explicit WxH. Downscale happens before quantization.

Both params are optional and independent — the user may pass one, both, or neither. Parse them out of natural phrasing too ("quality 60", "resize to 1600px wide", "under 500KB" → pick a quality that gets there).

Defaults

Unless told otherwise:

Quality: --quality=65-85 — visually near-lossless for screenshots and UI captures, typically 60-80% smaller.
Resolution: unchanged (no downscale).
Speed: --speed 1 (slowest, best compression — these are small files, it's fine).
Strip metadata: --strip.

Don't interrupt to confirm defaults. Only ask when the request is genuinely ambiguous (e.g. "make it tiny" with no size target).

Commands

Downscale first, only if resolution was given:

sips -Z <max-dimension> "<input>" --out "<work>.png"      # max dimension, preserves aspect
sips -z <H> <W> "<input>" --out "<work>.png"              # explicit HxW (note: height first)

Then quantize:

pngquant --quality=<min>-<max> --speed 1 --strip --force --output "<output>.png" -- "<input>.png"

Notes:

pngquant exits 99 when it can't hit the requested minimum quality and writes nothing. That's not a crash — widen the range (e.g. 45-85) and retry once, and say so in the report.
Exit 98 means the output would be larger than the input; keep the original in that case.
Write to a new file by default (<name>-min.png alongside the source). Only overwrite in place when the user explicitly asks; --force is there for re-runs, not for silently clobbering originals.
Non-PNG input (jpg, webp, heic): convert with sips -s format png first, or tell the user this skill is PNG-only if they want jpeg-native compression.
Multiple files: loop, and keep going if one fails — report failures at the end rather than aborting.

After running

Report a compact table: filename, before → after size, percent saved, and the final dimensions if you downscaled. Mention the quality range used, and flag any file that was skipped or needed a widened range.

---name: ps-image-compressiondescription: Compress PNG images with pngquant, optionally downscaling first, driven by a quality or resolution parameter. Use when the user says "compress this image", "make this png smaller", "optimize screenshots", "resize to 1200px", or invokes /ps-image-compression.---# Compress PNG ImagesShrink PNG files using `pngquant` (lossy palette quantization), with optional downscaling via `sips` before quantization. Requires `pngquant` (`brew install pngquant`); `sips` ships with macOS.## Inputs- **Target**: one or more PNG files, or a directory/glob. If the user didn't name one (or it doesn't exist), list PNGs in the current directory and ask them to pick.- **`quality`** (optional): target visual quality, given as a number `0-100` or a range like `65-85`. Maps to pngquant's `--quality=min-max`.- **`resolution`** (optional): max dimension in pixels (e.g. `1200`) or an explicit `WxH`. Downscale happens *before* quantization.Both params are optional and independent — the user may pass one, both, or neither. Parse them out of natural phrasing too ("quality 60", "resize to 1600px wide", "under 500KB" → pick a quality that gets there).## DefaultsUnless told otherwise:- Quality: `--quality=65-85` — visually near-lossless for screenshots and UI captures, typically 60-80% smaller.- Resolution: unchanged (no downscale).- Speed: `--speed 1` (slowest, best compression — these are small files, it's fine).- Strip metadata: `--strip`.Don't interrupt to confirm defaults. Only ask when the request is genuinely ambiguous (e.g. "make it tiny" with no size target).## CommandsDownscale first, only if `resolution` was given:```bashsips -Z <max-dimension> "<input>" --out "<work>.png" # max dimension, preserves aspectsips -z <H> <W> "<input>" --out "<work>.png" # explicit HxW (note: height first)```Then quantize:```bashpngquant --quality=<min>-<max> --speed 1 --strip --force --output "<output>.png" -- "<input>.png"```Notes:- pngquant **exits 99** when it can't hit the requested minimum quality and writes nothing. That's not a crash — widen the range (e.g. `45-85`) and retry once, and say so in the report.- Exit 98 means the output would be larger than the input; keep the original in that case.- Write to a new file by default (`<name>-min.png` alongside the source). Only overwrite in place when the user explicitly asks; `--force` is there for re-runs, not for silently clobbering originals.- Non-PNG input (jpg, webp, heic): convert with `sips -s format png` first, or tell the user this skill is PNG-only if they want jpeg-native compression.- Multiple files: loop, and keep going if one fails — report failures at the end rather than aborting.## After runningReport a compact table: filename, before → after size, percent saved, and the final dimensions if you downscaled. Mention the quality range used, and flag any file that was skipped or needed a widened range.

Frequently asked questions