Stable Diffusion Prompt Optimizer
Analyzes and enhances Stable Diffusion prompts using the Automatic1111 WebUI API /sdapi/v1/txt2img endpoint. Applies prompt weighting, negative prompt generation, and A1111-specific syntax like (keyword:weight).
npx skills add agentskillexchange/skills --skill stable-diffusion-prompt-optimizer
The Stable Diffusion Prompt Optimizer skill enhances text-to-image prompts for Automatic1111’s Stable Diffusion WebUI. It connects to the /sdapi/v1/txt2img REST endpoint and analyzes prompt effectiveness by generating test batches with varied parameters. The skill applies A1111-specific syntax enhancements including attention weighting with (keyword:1.3) notation, BREAK tokens for 75-token chunk boundaries, and alternating prompt syntax [word1|word2] for creative blending. Negative prompts are auto-generated based on the positive prompt content using an embedded quality-tag database covering common artifacts like extra fingers, blurry backgrounds, and watermarks. The skill queries /sdapi/v1/sd-models to match prompts with appropriate checkpoint models and uses /sdapi/v1/samplers to recommend optimal sampler settings. Batch comparison mode generates A/B test grids using the /sdapi/v1/img2img endpoint with controlled seed variation. LoRA trigger words are automatically detected and inserted via /sdapi/v1/loras endpoint enumeration. CFG scale and step count optimization uses binary search to find the quality-performance sweet spot.