stability-ai_skill

Overview

This skill generates high-quality images using Stability AI models (SDXL, SD3, Stable Image Core). It produces image files and accompanying JSON metadata, and supports presets, aspect ratios, and output formats for reproducible results. Designed to be run locally with a simple CLI and an API key.

How this skill works

The skill sends text prompts and generation parameters to the Stability AI API and saves the returned image bytes to disk. It also writes a metadata JSON record that includes prompt, negative prompt, model, parameters, timestamp, and API version. A CLI wrapper exposes options for model, steps, CFG scale, seed, style presets, aspect ratio, and output format.

When to use it

• You want to generate an image from a text prompt (e.g., "draw this", "make a picture").
• You need reproducible outputs via seeds and explicit generation parameters.
• You want multiple style presets and common aspect ratios without building a custom pipeline.
• You need local files plus detailed metadata for auditability or replay.
• You want quick experimentation with different models or advanced options like steps and CFG scale.

Best practices

• Set STABILITY_API_KEY in your environment before running the CLI to avoid runtime errors.
• Start with style presets and a seed to iterate quickly, then fine-tune steps and CFG for quality vs. speed.
• Use negative prompts to reduce unwanted artifacts or elements in the image.
• Choose output format according to use: PNG for quality, JPG for size, WebP for efficiency.
• Monitor disk usage; the skill auto-cleans older files but adjust retention if you need longer archives.

Example use cases

• Generate a product mockup in photographic style and save the image plus metadata for design review.
• Create concept art with specific aspect ratios (16:9 or 9:16) for social posts or video assets.
• Produce a reproducible series of variations using a fixed seed to compare models or parameters.
• Batch-generate themed images with a negative prompt to avoid common artifacts.
• Archive generated images and metadata for later auditing, training, or publication.