siliconflow-img-gen

What problem does it solve?

It helps you turn prompts into usable images (or edit existing images) without manually dealing with image model APIs, while ensuring the outputs are validated.

Core Features & Use Cases

• Text-to-image generation: Create images from a text prompt using a default Qwen image generation model, with a built-in fallback model when rate limits are hit.
• Image-editing from sources: Modify one to three source images by combining them with your prompt, using the default Qwen image-edit model when you provide the image URL(s).
• Local outputs + review flow: Downloads generated results to a timestamped output directory and produces prompts.json plus a simple index.html gallery for quick inspection; includes a required validation step to detect and retry abnormal “single-color” outputs.

Use case example: You need banner images for a product page—generate several prompt variations in text-to-image mode, then edit a selected image using a “make it look more like night mode” instruction while reviewing a gallery and re-trying any invalid generations.

Quick Start

Use the siliconflow-img-gen skill to generate an image by running the command: python3 {baseDir}/scripts/gen.py --prompt "a futuristic city at dusk".