What problem does it solve?
Provide a unified, configurable CLI workflow to generate and edit images from text prompts and reference images across multiple provider APIs, removing the friction of switching providers, resolving models, and managing output settings.
Core Features & Use Cases
- Multi-provider support: Works with Google (Gemini/Imagen), OpenAI (GPT Image), DashScope (阿里通义万象), and Replicate with consistent CLI behavior.
- Flexible inputs: Accepts direct prompts, prompt files, and reference images for edits or multimodal generation.
- Preferences & model management: Uses EXTEND.md and environment variables to persist default provider, model, quality, aspect ratio, and image size with clear precedence rules.
- Robust execution: Validates inputs, selects providers automatically, retries transient failures once, supports sequential default generation and optional parallel batches for large jobs.
- Use case: Quickly produce cover illustrations, social graphics, or batch art assets with consistent sizing and provider fallback.
Quick Start
Use the baoyu-image-gen skill to generate a 2k aspect 16:9 image from the prompt "A serene mountain lake at sunset" and save it as output.png.