siliconflow-img-gen

To generate images from text prompts using SiliconFlow, you run a Python script with your prompt string. The Skill uses the default Qwen image generation model to render your text into a PNG output and automatically saves it to a local directory for review.

Yes, you can edit existing images in SiliconFlow by providing one to three source image URLs along with your text prompt. The Skill uses the Qwen image-edit model to modify the reference images based on your instructions and outputs the resulting PNG file.

Yes, you must set the SILICONFLOW_API_KEY environment variable before using the image generation script. This key is required to authenticate your requests to the SiliconFlow Images API endpoint for rendering your prompts into PNG outputs.

If the image generation returns a blank or single-color output, the Skill automatically detects this invalid result and retries the rendering up to three times. This validation step ensures you receive a properly generated PNG image for your marketing visuals or creative workflows.

When the prompt-to-image conversion hits rate limits on the default Qwen model, the Skill automatically falls back to a secondary image generation model. This ensures your text prompts are still processed and output as PNG files without interrupting your creative workflow.

Generated images are downloaded to a timestamped output directory on your local system. The Skill also produces a prompts.json file and a simple index.html gallery, allowing you to quickly inspect and review the PNG outputs and their corresponding text prompts.