Z-Image Skill
https://github.com/yizhiyanhua-ai/zimage-skill- Category
- Developer Tools
- Rank
- No. 2011Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- yizhiyanhua-ai
- GitHub
- 52 stars
- Date
About
Claude Code skill that generates AI images from natural-language prompts.
What it does
It submits a text prompt to ModelScope’s Z-Image-Turbo service, polls the asynchronous job, downloads the first returned image, and saves it locally as JPG or PNG.
Why it's ranked here
The tool has a narrow, understandable design with few dependencies and useful bilingual guidance. Its main weakness is operational robustness: error handling is basic, polling is fixed, and the documentation contradicts the implementation about whether an API key is optional.
What's good
Setup needs only Python, two common packages, and a ModelScope token. Users can choose an output path, write prompts in Chinese or English, and receive a normal local image file. The small implementation is easy to inspect and modify.
Tradeoffs
Every generation depends on ModelScope network availability, service capacity, moderation, and credentials. Jobs time out after 120 seconds and poll every five seconds. The script takes the first returned image only, exposes no generation parameters beyond the prompt, and provides no retry strategy. It does not edit existing images.
How to use it well
It suits Claude Code users who need quick avatars, article illustrations, social graphics, memes, or rough design inspiration. Write detailed prompts with subject, style, lighting, and composition, then specify the destination and format. Choose another tool when you need image editing, parameter control, batches, or production-grade recovery.
Technical notes+
generate.py is a Python command-line script that reads MODELSCOPE_API_KEY, POSTs to https://api-inference.modelscope.cn/v1/images/generations with asynchronous mode enabled, polls the task endpoint every five seconds, and saves the first output through Pillow. requirements.txt declares requests>=2.25.0 and Pillow>=8.0.0. SKILL.md exposes the script as a Claude Code skill with Bash, Write, and Read permissions. Its statement that the API key is optional conflicts with generate.py, which exits when the variable is absent. README.md documents natural-language installation, JPG and PNG output, Chinese and English prompts, and a 120-second timeout.
Observed
- License
- MIT License
- Primary language
- Python
- Install surface
- Claude Code skill installed from the GitHub repository, with Python dependencies installed separately
- Interfaces
- Claude Code skill and positional command-line script
- Dependencies
- requests 2.25.0 or newer and Pillow 8.0.0 or newer
- Service
- ModelScope API using Tongyi-MAI/Z-Image-Turbo
- Prompt languages
- Chinese and English
- Output formats
- JPG by default, with PNG supported through the output filename
Read from README.md, requirements.txt, generate.py, LICENSE, SKILL.md, README_CN.md.
What it can do
Generate AI images from text descriptions
Natural language text prompts → AI-generated images
Create visual content from written concepts
Descriptive text about desired image → Digital image file
Transform textual ideas into visual representations
Text-based image specifications → Rendered image
Process natural language for image generation
User-provided text prompts → Corresponding visual artwork
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.