generate-images - SKILL.md Agent Skill

name: generate-images description: Workflow management for Imagen 4 image generation and post-processing. Use when generating AI assets, enforcing aspect ratios (e.g. 16:9), changing resolutions (1K or 2K), or optimizing file sizes for web delivery.

Generate Images Skill

This skill optimizes the workflow for using the gemini-3.1-flash-image model to generate and process visual assets for Liferay projects.

Core Workflows

Image Generation: Efficiently utilize the Gemini 3.1 Flash Image model with descriptive prompts via Python REST API calls.
Aspect Ratio Enforcement: Use the API's aspectRatio parameter natively or use Python (Pillow) to crop generated images into specific formats.
Resolution Control: Change resolution sizes using the sampleImageSize parameter (prefer 1K over 2K).
Format Optimization: Convert large PNG assets to compressed JPEGs or WebP for better web performance.
Icon & Pattern Generation: Specialized instructions for UI elements and tiling backgrounds.

STRICT EXECUTION PROTOCOL (MANDATORY READS)

You MUST NOT attempt to write API calls to the model from memory. You MUST use the read_file tool to read the following reference document BEFORE generating images:

Gemini Flash Image API Usage: You MUST read GEMINI_3_1_FLASH_IMAGE_GUIDE.md to obtain the correct Python script templates, endpoint URLs, and valid parameter configurations (e.g., aspectRatio, sampleImageSize).