gbro-cover-design

Feed an article to an AI agent. It reads the content, asks you three questions, and outputs a cover prompt that can be directly used with image generation tools. The title is automatically generated, and all composition logic is built-in. The aspect ratio is fixed to 3:4 vertical.

Works with Claude Code, Codex, and any AI agent that supports custom skills.

Preview

The following covers were generated using this skill:

Use Cases

• Content creators and social media operators who need to quickly generate cover images for articles
• Users with no design or prompt engineering experience who want professional covers effortlessly
• Anyone who needs to produce a consistent set of covers in bulk

Core Capabilities

• Automatic content analysis: The AI reads your article and extracts key information for cover design
• Three-round questioning: Guides you through composition style, materials, and visual details step by step
• 10 composition styles: Recommended based on content, with manual selection also available
• Fixed 3:4 vertical aspect ratio: Ideal for social media covers, with built-in safe area definitions
• Face consistency support: Uses a reference photo to maintain facial identity (requires a model that supports multi-reference image input)
• Built-in example prompt library: 8 complete example prompts are provided in the repository (references/examples.md)
• Automatic title generation: The skill generates a cover title based on your article

How It Works

1. You provide: Send the article text to the AI agent.
2. Skill activates automatically: gbro-cover-design reads the article and starts three rounds of questions:
   • Round 1: Recommends a composition style based on the article and asks for the cover title.
   • Round 2: Asks for a face reference photo, product screenshots, or other materials.
   • Round 3: Lists options for expression, background color, font, and font color in one go; anything not mentioned is left to the model.
3. Output: A prompt string that can be directly used with image generation tools, containing all composition elements, face references, and text layout.
4. You generate the image: Copy the prompt into a supported image generation tool (e.g., Jimeng, Seedream 4.0, Nano Banana, GPT-Image) to create the cover.

What You Need to Provide

• Article content: The body or summary of the article you want a cover for.
• Face reference photo (optional): A clear front-facing photo. On first use you can save it to assets/my-face.png in the skill directory as a default. You can also upload a new one each time.
• Other materials (optional): Product images, UI screenshots, brand assets, etc., to include in the cover.
• Image generation model: Must support multi-reference image input to maintain face consistency.

What You Get

• A cover prompt string ready for image generation tools, including:
  • Composition style and layout
  • Title text
  • Placement of people, products, or other materials
  • Color, font, expression, and other details
• No design knowledge or prompt engineering skills required.

10 Composition Styles

Full prompt templates for each style are available in the repository at references/style-XX-*.md.

Notes

• First-time setup: The first time the skill is triggered, it will guide you through configuring a face reference photo and confirming the image generation model (no API key required). The configuration is saved in config.md and won't be repeated.
• Face consistency: Requires a model that supports multi-reference image input (e.g., Jimeng/Seedream 4.0, Nano Banana, GPT-Image). Otherwise, facial identity cannot be guaranteed.
• Fixed aspect ratio: The output prompt is always 3:4 vertical, suitable for most social media cover formats.
• Prompt only: This skill only generates prompt text. You must use a separate image generation tool to produce the actual cover.
• Template completeness: The skill relies on the references/ directory for style templates and example prompts. Make sure it is fully included (Skill Hub installations will include it automatically).