Core Concept: Analyze Style → Design Prompts → User Confirmation → Generate Images, Avoid Credit Wastage

• Text-to-Image: Generate images based on text descriptions
• Image-to-Image: Generate images with similar styles based on reference images
• Batch Generation: Generate multiple images in batches using prompt templates
• Consistent Style: Maintain unified visual style for a series of images

• Long Image Merging: Vertically stitch a series of images into a single long image
• PPT Packaging: Package a series of images into a PPT file (one image per page)
• Image Hosting Management: Upload images to hosting services and manage deletion links

• Users explicitly request "generate images", "create illustrations", "make artwork"
• Users provide text descriptions and need visual representation
• Users provide reference images and request images with similar styles
• Users need to create illustrations for articles, PPTs, or social media
• Users need to batch generate a series of images with consistent style
• Users need to merge multiple images into a long image or PPT

• Users only ask how to generate images (provide advice only)
• Users need to edit existing images (use image editing tools)
• Users need image recognition or analysis (use visual analysis tools)
• Users need screenshots or screen recording (use system tools)

~/.claude/skills/pw-image-generation/scripts/

1. Determine the directory path of this SKILL.md file as

   SKILL_DIR

2. Script path =

   ${SKILL_DIR}/scripts/<script-name>.ts

3. Replace all

   ${SKILL_DIR}

   in this document with the actual path

generate-image.ts

[output directory]

upload-image.ts

<image path>

delete-image.ts

list|delete <index>|delete-all

merge-to-long-image.ts

<image directory> <output file>

merge-to-pptx.ts

<image directory> <output file>

analyze-image.ts

<image URL or path>

npx -y bun ~/.claude/skills/pw-image-generation/scripts/generate-image.ts [output directory]

• [output directory]

  : Optional, specify the image save directory, default is

  ./images

1. Read all prompt files in the

   prompts/

   directory
2. Display each prompt content one by one
3. Ask whether to generate the image (y/n/q)
4. Generate and save the image to the output directory
5. Support interruption and resumption

• y

  (yes): Generate current image

• n

  (no): Skip current image and continue to the next

• q

  (quit): Exit the generation process

undefined

npx -y bun ~/.claude/skills/pw-image-generation/scripts/upload-image.ts <image path>

• <image path>

  : Required, path to the local image file to upload

• Automatically upload to freeimage.host (permanent storage)
• Return usable image URL
• Automatically save deletion link to

  .upload-history.json

• Used for reference image upload in image-to-image generation

undefined

npx -y bun ~/.claude/skills/pw-image-generation/scripts/merge-to-long-image.ts <image directory> <output file>

• <image directory>

  : Required, directory containing images to merge

• <output file>

  : Required, filename of the output long image

• Vertically stitch multiple images
• Automatically recognize jpg/png/gif/webp formats
• Sort by filename numbers
• Maintain original image width, automatically calculate height

brew install imagemagick

undefined

npx -y bun ~/.claude/skills/pw-image-generation/scripts/merge-to-pptx.ts <image directory> <output file>

• <image directory>

  : Required, directory containing images to package

• <output file>

  : Required, filename of the output PPT

• One image per page
• Automatically recognize jpg/png/gif/webp formats
• Sort by filename numbers
• 16:9 aspect ratio, automatically adapt to page size
• Images displayed in center

undefined

• API key not configured or invalid
• Incorrect prompt format
• Network connection issues
• Insufficient API credits

undefined

• Image file does not exist or incorrect path
• Unsupported image format
• Image file too large (exceeds 10MB)
• Network connection issues

undefined

• ImageMagick not installed
• ImageMagick not added to PATH

undefined

• Image directory is empty or does not exist
• Unsupported image format
• Invalid output file path

undefined

• Prompt descriptions are not specific enough
• Mixed Chinese and English in prompts
• Prompts are too long or too short
• Style descriptions are unclear

undefined

• Unstable network
• Manual interruption by user (Ctrl+C)
• API rate limiting

• The script will automatically skip already generated images
• Re-run the generation command to continue
• Already generated images will not be regenerated
• Use

  n

  option to skip unwanted images

• Incorrect configuration file path
• Incorrect configuration file format
• Configuration template not copied

undefined

• Xiaohongshu also supports 1:1 (1080×1080) and 4:3 (1440×1080)
• WeChat Official Account secondary cover can use 200×200 (1:1)

• Specify size in prompts:

  size: 1920x1080

  or

  aspect_ratio: 16:9

• Prioritize recommended sizes to avoid composition imbalance caused by later cropping
• Different platforms have different size requirements, confirm target platform before generation

• Specific and clear: Detailed description of scene, style, color, and composition
• Consistent style: Use preset style descriptions from the style library
• Consistent language: All English or all Chinese, avoid mixing
• Moderate length: 50-200 words is appropriate; too short lacks details, too long is confusing

[Style] + [Subject] + [Scene] + [Atmosphere] + [Technical Details]

Example:
Watercolor style, a cute kitten sitting on a windowsill, warm afternoon sunlight, soft colors, hand-painted texture

• references/style-library.md

  : 9 preset styles

• references/prompt-templates/提示词模板.md

  : Prompt templates

1. First generate 1-2 test images to verify prompt effectiveness
2. Adjust prompts based on test results
3. Batch generate remaining images after confirming satisfactory results
4. Use confirmation mechanism to avoid wasting API credits
5. Regularly back up generated images

• Carefully check prompts before each generation
• Use

  n

  option to skip unwanted images
• Use

  q

  option to exit promptly
• Do not blindly batch generate

my-image-project/
├── config/              # Configuration files (not submitted to git)
├── template/            # Reference images
├── prompts/             # Prompt files
│   ├── 01-封面.md
│   ├── 02-内容1.md
│   └── 03-内容2.md
├── images/              # Generated images
│   ├── 01-封面.png
│   ├── 02-内容1.png
│   └── 03-内容2.png
└── output/              # Final output (long images/PPTs)

• Prompt files: Use numeric prefixes for sorting (01-, 02-, 03-)
• Generated images: Automatically use prompt file names
• Facilitates batch processing and sorting

• Use smaller image sizes (1024x1024 instead of 2048x2048)
• Avoid overly complex prompts
• Use image-to-image appropriately (faster than text-to-image)

• Use confirmation mechanism, do not skip confirmation
• Test single images first, then batch generate
• Save good prompts as templates for reuse
• Regularly clean up unnecessary images on hosting service

• Prompt descriptions are clear and specific
• Style descriptions are unified and consistent
• Test image results are satisfactory
• Image size and format are correct
• File naming is standardized and organized
• Important images are backed up

• Inconsistent image style: Check style descriptions in prompts
• Poor image quality: Add technical detail descriptions (e.g., "high definition", "rich details")
• Image content deviation: Make prompts more specific and add constraint conditions

• Asks for confirmation before each generation
• Supports skipping already generated images
• Generate one image at a time to avoid batch credit consumption

• Configuration file

  config/secrets.md

  is optional
• Uses default configuration when not configured (requires setting environment variable API_KEY)
• Configuration files are not submitted to version control (already in .gitignore)
• Supports custom API endpoints and models

config.example/secrets.md

• API_BASE_URL

  : API base URL (default: OpenAI API)

• ANALYSIS_MODEL_ID

  : Image analysis model (default: gpt-4-vision-preview)

• GENERATION_MODEL_ID

  : Image generation model (default: dall-e-3)

• API_KEY

  : API key (required)

undefined

• Watercolor (watercolor) - Soft and warm
• Flat Design (flat-design) - Modern and concise
• 3D Render (3d-render) - Stereoscopic and realistic
• Oil Painting (oil-painting) - Artistic and classic
• Cyberpunk (cyberpunk) - Sci-fi and futuristic
• Pixel Art (pixel-art) - Retro and nostalgic
• Hand-Drawn Illustration (hand-drawn) - Warm and personalized
• Photorealistic (photorealistic) - Highly realistic
• Abstract Art (abstract) - Emotional expression

references/style-library.md

1. Confirm Requirements: Ask users about image usage, style preferences, and quantity
2. Prepare Environment: Guide users to create project directories and configurations
3. Design Prompts: Design prompts based on requirements, refer to the style library
4. Generate Tests: First generate 1-2 test images
5. Batch Generation: Batch generate after confirming satisfactory results
6. Post-Processing: Merge into long images or package into PPT as needed