guanyang/sticker-factory

Sticker Factory (Auto-Generator & Packer)

This skill serves as an Automated Production Line that orchestrates the entire lifecycle of sticker creation: from AI generation to WeChat-compliant asset packaging.

Execution Workflow

Step 1: Generate Sticker Grid

Adhere to the Universal Grid Logic defined in skills/grid-sticker-generator/SKILL.md.

• Prompt Construction:

  • Core: User's described character/theme.
  • Format Constraints (CRITICAL): Inherit all Visual Design Standards from grid-sticker-generator (4x4 Grid, No Text, Flat Vector).
  • Content Logic:
    • Default: Load the Office Worker Template (references/office_worker_template.md) from the generator skill to fill slots with office memes.
    • Custom: If user specifies a different theme, generate 16 varied actions relevant to that theme satisfying the grid constraints.

• Action:

  • Call generate_image with the constructed prompt.

Step 2: Auto-Process with Sticker Maker

Once the image is successfully generated, immediately trigger the packaging robot.

• Command:

  ./skills/wechat-sticker-maker/scripts/run.sh "PATH_TO_GENERATED_IMAGE" --output "OUTPUT_DIR" --layout 4x4
  

  • PATH_TO_GENERATED_IMAGE: The absolute path returned by the generate_image tool.
  • OUTPUT_DIR: If user specified a custom path, use it. Otherwise, default to output/stickers_[theme].

• Metadata Auto-Population (CRITICAL): Before moving to the next step, the Agent must:

  • Automatically edit and fill in info.txt and meta.txt with actual names, descriptions, and sticker meaning words, replacing the template placeholders, based on the theme generated.
  • Copy the original generated grid image into the output folder (e.g. as source.png or grid_image.png).

Step 3: Generate Matching WeChat Sticker Assets

Orchestrate the design of matching promotional materials based on the generated IP character using skills/wechat-sticker-assets-designer/SKILL.md.

• Action:
  1. Analyze IP: Extract visual features (hair, colors, outfit) from the generated grid image.
  2. Generate Assets:
     • Banner: Aspect ratio 15:8 (750x400 target). Rich story-rich scene, no text, no white/transparent background.
     • Solicitation Image: Aspect ratio 4:3 (750x560 target). Begging pose, including text "您的赞赏是我们最大的动力！", non-white background.
     • Acknowledgment Image: Aspect ratio 1:1 (750x750 target). Celebratory pose, including text "感谢您的慷慨赞赏！", non-white background.
  3. Crop & Resize: Run the conditional crop and resize wrapper script on each generated asset to achieve the exact WeChat specification sizes:

     ./skills/wechat-sticker-assets-designer/scripts/run.sh "PATH_TO_BANNER" --width 750 --height 400 --anchor center
     ./skills/wechat-sticker-assets-designer/scripts/run.sh "PATH_TO_SOLICITATION" --width 750 --height 560 --anchor top
     ./skills/wechat-sticker-assets-designer/scripts/run.sh "PATH_TO_ACKNOWLEDGMENT" --width 750 --height 750 --anchor center
     

     • 保存路径 (CRITICAL): 将裁剪缩放后的三张配套图保存到该表情包输出目录下的 assets/ 子目录中（即 OUTPUT_DIR/assets/），严禁零散放置在外部。

Step 4: Final Delivery

Report the complete production results to the user:

1. Preview: Show the generated grid sheet and the matching promotional assets (Banner, Solicitation, Acknowledgment).
2. Asset Package: Provide the absolute clickable link to the consolidated output directory:
   • Stickers & Assets Project Folder: OUTPUT_DIR (Stickers inside the root folder, and Banner/Solicitation/Acknowledgment assets inside the assets/ subfolder).

Example Command

"Generate a set of generic 'Cute Robot' stickers and assets under a single folder." "Production line: Make me an office worker set and matching assets for a 'Tired T-Rex'."