Image Compression Skill

Compress PNG images using Python/Pillow. Two modes:

• Lossless — no quality loss, but moderate size reduction (typically 1-30%)
• Lossy — reduces color count to 256, significant size reduction (typically 50-80%) with minimal visual difference

Supports both single file and batch (folder) processing.

When the user provides input

They may say things like:

• "Compress this image" — ask if they want lossless or lossy
• "Compress all PNGs in this folder" — batch process a directory
• "Make this image smaller" — default to lossy for significant savings
• Point to a file or folder path

If the user doesn't specify mode, ask: "lossless (keep quality) or lossy (much smaller)?"

Workflow

1. Understand the scope: Single file or batch? Lossless or lossy?

1. Lossless:
• Save with optimize=True and compress_level=9
• Best for screenshots with text/UI elements where sharpness matters

1. Lossy:
• Quantize to 256 colors using Image.Quantize.MEDIANCUT
• Save with optimize=True
• Best for photos, gradients, or any image where small file size is priority

1. Report results: original size → compressed size → savings percentage

Script

A helper script compress_png.py handles both modes:

# Lossless (default)
python scripts/compress_png.py <path> [--output OUTPUT] [--overwrite]

# Lossy
python scripts/compress_png.py <path> [--output OUTPUT] [--overwrite] --lossy

• : Path to a PNG file or a directory
• --output: Output path (single file) or output directory (batch)
• --overwrite: Overwrite original files
• --lossy: Use lossy compression (256 colors)

Example interaction

User: "compress this image C:\Users\me\image.png it's too big"

You: "Lossless or lossy?"

User: "lossy"

You: Compress, report:

> Original: 1.2 MB → Lossy: 440 KB (saved 63%)