Image

• You need to resize, crop, convert, or compress an image and the output must meet specific quality, format, or dimension requirements for a real delivery context (web, social, ecommerce, print).
• You are preparing visual assets for a platform with strict rules — marketplace product photos, social media safe zones, app store artwork, or web performance budgets.
• You need to inspect an image before editing: checking dimensions, aspect ratio, color profile, EXIF metadata, transparency, or compression damage.
• You are batch-processing a set of images and need consistent, safe rules applied across the entire set without risking the originals.

• You need to generate new images with AI — use the image-generation skill instead.
• You need deep vector graphics authoring or manipulation — use the svg skill.
• You need non-destructive photo editing, RAW processing, or print color work at a professional level — use the photography skill.

The skill routes decisions by delivery context — web, social, ecommerce, print, branding, screenshots, or accessibility — rather than applying generic defaults. Specialized sub-files (web.md, social.md, ecommerce.md, etc.) are loaded only when the exact context is known, preventing common format and crop mistakes.

Photos, screenshots, logos, diagrams, and social cards each have different format needs. The skill explicitly maps asset types to starting formats (AVIF/WebP for web photos, PNG for UI captures, SVG for logos) and flags edge cases like transparency, animated output, and aggressive platform recompression.

The skill enforces the right operation sequence: decide aspect ratio first, crop second, resize third, compress last. It guards against common mistakes like upscaling by default, over-eager 3x Retina exports, and crops that cut off faces, products, or critical UI elements.

EXIF orientation, GPS data, copyright fields, and color ICC profiles are treated as real delivery concerns, not afterthoughts. The skill provides explicit guidance on when to strip vs. preserve metadata based on destination (public web vs. editorial vs. archive).

Built-in web delivery defaults give agents a concrete forcing function: hero images under 200 KB, content images under 100 KB, thumbnails under 30 KB, raster icons under 5 KB. These are starting points that balance efficiency against visual quality and platform recompression headroom.

The skill enforces working from originals or clean masters, keeping per-destination exports separate, and spot-checking representative files before running a full batch. One wrong crop preset or lossy export can damage an entire set — this skill helps prevent that.

An agent preparing hero images and content photos for a website uses the skill to pick AVIF/WebP formats, hit file-size budgets, strip unnecessary EXIF data, and reserve layout dimensions to avoid CLS — then validates the result in the actual page context, not just locally.

Before uploading to a marketplace, an agent uses the skill to ensure product photos have clean white backgrounds, correct square-safe crops that protect the product subject, sufficient zoom detail, and consistent sizing across an entire catalog batch.

An agent producing banners, preview cards, and story assets for multiple platforms uses the skill to apply platform-specific dimensions and safe-zone rules, avoiding unsafe crops that cut off text or faces when feeds display previews differently than full posts.

An agent preparing UI screenshots for release notes or tutorials uses the skill to output lossless PNG (avoiding JPEG blur on sharp text), redact sensitive data still visible in captures, and size images correctly for documentation layouts.