prompt-zs

What This Skill Does

Prompt-ZS is a specialized OpenClaw AI agent skill designed to bridge the gap between abstract user intent and concrete AI-generated media. Acting as a professional prompt engineer, the agent dynamically switches between five distinct modes to handle AI image generation (Midjourney, Stable Diffusion), video creation (Runway, Pika), and professional translation. By employing a structured formula for prompt construction—covering elements like subject, environment, lighting, camera parameters, and physical constraints—it ensures that the output is highly optimized for specific AI model engines. The system strictly follows a JSON-first output format for machine-readable tasks and natural language generation for video descriptions, ensuring efficiency and precision without conversational overhead.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/bozoyan/prompt-zs

Use Cases

• Graphic Designers: Quickly generate complex image prompts using specific photographic parameters like lens, ISO, and aperture for high-fidelity renders.
• Video Content Creators: Define motion sequences for text-to-video tools, adhering to strict 5-second duration limits and physical motion logic.
• Multilingual Teams: Seamlessly bridge the gap between Chinese requirements and English-dominated AI model tokenization through high-precision translations.
• Asset Generation Workflows: Standardize creative output by using consistent tag-based structures for batch generation in Stable Diffusion.

Example Prompts

1. "生成提示词2：拍摄一张赛博朋克风格的咖啡馆照片，要求使用Sony A7R4和50mm镜头，光线昏暗，霓虹灯倒影清晰。"
2. "文生视频：生成一个宇航员在月球表面缓慢行走的5秒短视频，背景是浩瀚星空，镜头采用推拉效果。"
3. "基于图片生成视频：这张图片是平静的湖面，请让它动起来，产生微风拂过水面的涟漪效果。"

Tips & Limitations

• Strict Formatting: The skill prioritizes JSON output for image prompts. Ensure your consuming application is ready to parse structured data rather than raw text.
• Constraint Awareness: When generating video prompts, the agent is constrained by a 500-character limit to ensure prompt clarity and model compatibility. Avoid adding complex narrative plots that exceed the 5-second execution capability.
• Trigger Efficiency: The skill utilizes a fast-decision logic tree. By using keywords like '精准' or '视频' in your initial prompt, you ensure the agent selects the optimal processing mode immediately without further clarification. The skill is designed to be concise; do not expect conversational fillers like 'I am ready to help' as these are suppressed by the system's design constraints.