From gemini-images
Generate images from text, transform existing images with natural language, and combine multiple image references for visual consistency. Use when creating images, editing visuals, or maintaining consistent characters/objects across multiple generations. Supports reference sheets, structured prompts, and visual anchor repetition.
How this skill is triggered — by the user, by Claude, or both
Slash command
/gemini-images:gemini-imagesThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
CLI: `npx @krasnoperov/gemini-images <command> [args] [options]`
CLI: npx @krasnoperov/gemini-images <command> [args] [options]
Three operations that compose into any workflow:
generate "<prompt>" # Text → Image
edit <image> "<prompt>" # Image + Instructions → Image
compose <img1> <img2> ... "<prompt>" # Images + Instructions → Image
export GEMINI_API_KEY="your-key" # Get at https://aistudio.google.com/app/apikey
npx @krasnoperov/gemini-images generate "pixel art tree, white background" --output tree.png
npx @krasnoperov/gemini-images edit tree.png "add glowing runes" --output tree-magic.png
npx @krasnoperov/gemini-images compose hero.png sword.png "character holding sword" --output hero-armed.png
See examples/ for complete working scripts.
Note: Output extension may differ from requested (e.g., .webp instead of .png) based on Gemini's response. Match by filename pattern when chaining operations.
Key Insight: Gemini has spatial understanding - it generates different views/poses while preserving visual features when given specific references.
Image 1:, Scene:, Character:, Lighting:Image 1: [description]
Image 2: [description]
Scene: [setting]
Character: [entity] from image [N], [pose]
Lighting: [direction, quality]
Camera: [angle, shot type]
Constraints: [what to preserve]
--model <model> gemini-3-pro-image-preview (default) or gemini-2.5-flash-image
--aspect-ratio <ratio> 1:1, 16:9, 9:16, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 21:9
--image-size <size> 1K (default), 2K, 4K (Pro required for 2K/4K)
--output <path> File or directory (default: ./output/)
gemini-3-pro-image-preview:
gemini-2.5-flash-image:
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub krasnoperov/claude-plugins --plugin gemini-images