qwen-image

What This Skill Does

The qwen-image skill is a robust interface for the Alibaba Cloud DashScope Qwen Image API, commonly known as Tongyi Wanxiang. It empowers OpenClaw to generate high-fidelity, professional-grade images from natural language text descriptions. Whether the user provides prompts in English or Chinese, the skill handles complex visual synthesis with advanced models like qwen-image-max and qwen-image-plus. It integrates seamlessly into the OpenClaw workflow by returning image URLs directly, which are then rendered into the chat interface as Markdown content. This ensures a fast, clean user experience without the need for immediate local storage unless explicitly requested.

Installation

To install this skill, use the OpenClaw CLI tool in your terminal:

clawhub install openclaw/skills/skills/robin797860/qwen-image

Once installed, ensure your authentication is configured. The skill automatically searches for your API key in several locations: the global OpenClaw configuration file (~/.openclaw/openclaw.json) under both models.providers.bailian.apiKey and skills."qwen-image".apiKey, or via the DASHSCOPE_API_KEY environment variable. You can obtain your unique key from the official DashScope console.

Use Cases

This skill is ideal for creators, developers, and researchers who require:

1. Concept Visualization: Quickly iterating on design ideas by turning text into visual concepts.
2. Multilingual Content: Generating culturally accurate imagery using native Chinese prompt support.
3. Content Marketing: Producing bespoke social media assets or illustrations.
4. Prototyping: Rapidly generating high-quality placeholders for web or application development.

Example Prompts

• "生成一张未来赛博朋克风格的上海滩街道图片，强调霓虹灯和雨水反射，使用16:9比例"
• "Generate a high-quality oil painting of a peaceful mountain landscape during autumn, use the max model for best quality."
• "Create an image of a minimalist workspace with a laptop and a succulent, square format please."

Tips & Limitations

• Prompting: While the model is powerful, adding descriptive details (lighting, style, atmosphere) yields significantly better results.
• Aspect Ratios: Utilize the --size flag to match your desired output format, such as 1664*928 for cinematic landscape or 1024*1024 for square.
• Performance: For best results, allow the automatic prompt enhancement feature (default) to assist in refining your input. If you have a specific artistic vision, use the --no-prompt-extend flag.
• Storage: Remember that image URLs are temporary. If you need to keep a specific image, download it manually or use a script to capture the URL output immediately after generation.