| name | zimage |
| description | Use when the user asks to generate images via ModelScope Z-Image API (Tongyi-MAI/Z-Image-Turbo); run the bundled Node.js script and require MODELSCOPE_API_TOKEN for live calls. |
Z-Image Generation Skill
This skill generates an image from a natural-language prompt using the ModelScope Z-Image API.
When to use
- Generate a new image given a prompt
- Specify size (width × height)
- Control diffusion steps
Workflow
- Decide if the user wants an image and/or parameters.
- Collect required inputs:
- prompt
- width
- height
- steps
- negative prompt (optional)
- Run the bundled Node.js script (
index.js) with provided inputs.
- Return the result (image URL or base64 payload) to the user.
Default behavior
- If width or height is omitted, defaults to 1024.
- If steps is omitted, defaults to 30.
- Negative prompt is optional and passed as
negative_prompt in API payload.
- Requires
MODELSCOPE_API_TOKEN to be set in environment.
Notes
- Model:
Tongyi-MAI/Z-Image-Turbo
- API:
https://api-inference.modelscope.ai/v1/images/generations
- Free tier and rate limits depend on ModelScope account and current policy.
CLI Example
node index.js "a cat portrait" 512 512 30 "blurry, low quality, deformed"
