image-process

Image Process

When to Use

Use this skill when the user already has local image files and needs deterministic post-processing:

• Resize, crop, thumbnail, rotate, convert format, compress, set quality, or strip EXIF.
• Normalize a batch of images for web, social, app, marketplace, or document delivery.
• Produce non-destructive output files while preserving originals.

Do not use this skill for semantic AI edits such as "remove this person", "make the subject prettier", "change the background to a beach", or "generate a new image". Those require image design/editing or an AI image tool.

Workflow

1. Confirm inputs:
   • exact file paths or glob range.
   • target operation, dimensions, output format, output directory, and whether metadata should be stripped.
2. Protect originals:
   • default to writing into an output directory or adding a suffix.
   • do not overwrite originals unless the user explicitly confirms overwrite.
3. For batches:
   • list or count matched files before processing.
   • confirm if the operation is destructive, lossy, or large-scale.
4. Process in a stable order:
   • auto-orient from EXIF, crop/fit, resize, convert, strip metadata, save quality.
5. Verify outputs:
   • path, dimensions, format, file size, and failures.

Script

Use scripts/image_process.py when Python and Pillow are available.

Examples:

python scripts/image_process.py --input cover.png --output-dir out --width 1080 --format webp --quality 85 --strip-exif
python scripts/image_process.py --input "images/*.png" --output-dir out --width 1080 --height 1080 --fit cover --format jpg --quality 90

Install dependency if needed:

python -m pip install Pillow

Do not install dependencies without user approval in restricted or shared environments.

Return Format

Return:

{
  "ok": true,
  "outputs": [
    {
      "path": "output.webp",
      "dimensions": "1080x1080",
      "format": "WEBP",
      "file_size": 123456,
      "operations": ["resize", "convert", "strip_exif"]
    }
  ],
  "failed": []
}