Stable Diffusion XL Pipeline
Orchestrates SDXL image generation via the Stability AI REST API with ControlNet conditioning, IP-Adapter style transfer, and automatic prompt enhancement using CLIP interrogation.
npx skills add agentskillexchange/skills --skill stable-diffusion-xl-pipeline
The Stable Diffusion XL Pipeline skill manages end-to-end image generation workflows using the Stability AI platform API. It handles text-to-image, image-to-image, and inpainting modes with full control over SDXL parameters including cfg_scale, steps, sampler selection, and seed management.
What this skill actually does
ControlNet integration supports Canny edge, depth map, and OpenPose conditioning for precise compositional control. The skill preprocesses input images using OpenCV for edge detection and MiDaS for depth estimation before sending to the API. IP-Adapter style transfer allows blending reference image aesthetics with text prompts.
The prompt engineering module uses CLIP interrogation to analyze reference images and generate optimized prompts. It includes a negative prompt library, automatic prompt weighting syntax, and A1111-compatible embedding triggers. Output images are post-processed with Real-ESRGAN upscaling and face restoration via GFPGAN, with metadata embedded as EXIF and PNG tEXt chunks.