Puzzle Image Generator

Create puzzle-ready background image prompts and generate images via the Gemini image preview API. Use when you need a clean puzzle base image (no text, no puzzle lines) and want a Node.js script to call `generateContent` and save the image to disk.

doggy8088 Updated Sep 17, 2026

Integrations & APIs

Install

Press C to copy

$ npx skillmds@latest add doggy8088/puzzle-image-generator

Installs into ~/.claude/skills (or .claude/skills inside a project). Restart Claude Code to pick it up.

＋ Add to collection

README badge

[![SkillMD](https://skillmd.com/badge/doggy8088/puzzle-image-generator.svg)](https://skillmd.com/skills/doggy8088/puzzle-image-generator)

Trust & safety

Audits pending

Not yet audited · commit 3821a00

Works with: Claude Code Claude.ai OpenAI Codex +

File contents

Puzzle Image Generator

Overview

Generate a puzzle-ready prompt and render a square (1:1) image using the bundled Node.js script.

Workflow

1. Write the prompt

Use this template and replace the bracketed parts. Keep it concise and literal.

Beautiful [scene subject] in [style], rich detail, balanced composition, high-contrast edges,
clean background, no text, no logos, no watermarks, no borders, no UI, no puzzle lines,
no grid, no segmentation, no labels, no typography. --AR 1:1

Prompt tips

Prefer clear shapes and distinct regions (helps puzzle piece separation).
Avoid tiny repeated patterns, heavy blur, or flat gradients.
Keep the image self-contained (no frames or letterboxing).
AI can freely choose any one scene subject and any one style from the 50-option lists below, then combine them in the prompt template.

Scene subject options (50)

Style options (50)

2. Generate the image

Run the script with your prompt and optional output folder. If no output folder is provided, the image is saved under images.

node scripts/generate_puzzle_image.js --prompt "YOUR_PROMPT_HERE" --out "optional/output/folder" --name "my-image.png"

Use this default numbering logic for --name argument:

Scan images/generated-*.png.
Extract numeric suffixes from matching filenames.
Compute the next N as max + 1 (default to 1 if no matches exist).
Generate with filename images/generated-{N}.png.

3. Post-image-generation rule

After generating images, derive N from current generated files before updating code:

Scan images/generated-*.png.
Extract numeric suffixes and set N to the maximum suffix (example: latest is generated-14.png => N = 14).
Replace hard-coded magic numbers with N in the index.html file, for example:
```
const GENERATED_NUMBERS = 15; // Replace 15 with N
```

Script Notes

Requires GEMINI_API_KEY in the environment.
Expects the Gemini response to include inlineData.data (base64 image).
Saves a .png by default and derives the extension from inlineData.mimeType when available.

---name: puzzle-image-generatordescription: Create puzzle-ready background image prompts and generate images via the Gemini image preview API. Use when you need a clean puzzle base image (no text, no puzzle lines) and want a Node.js script to call `generateContent` and save the image to disk.---# Puzzle Image Generator## OverviewGenerate a puzzle-ready prompt and render a square (1:1) image using the bundled Node.js script.## Workflow### 1. Write the promptUse this template and replace the bracketed parts. Keep it concise and literal.```textBeautiful [scene subject] in [style], rich detail, balanced composition, high-contrast edges,clean background, no text, no logos, no watermarks, no borders, no UI, no puzzle lines,no grid, no segmentation, no labels, no typography. --AR 1:1```**Prompt tips**- Prefer clear shapes and distinct regions (helps puzzle piece separation).- Avoid tiny repeated patterns, heavy blur, or flat gradients.- Keep the image self-contained (no frames or letterboxing).- AI can freely choose any one scene subject and any one style from the 50-option lists below, then combine them in the prompt template.**Scene subject options (50)**1. Snowy mountain cabin2. Sunset beach boardwalk3. Quiet forest waterfall4. Desert cactus valley5. Lavender flower field6. Tropical coral reef7. Cozy bookstore interior8. Vintage train station9. Bustling night market10. Medieval castle courtyard11. Japanese temple garden12. Colorful hot air balloons13. Autumn maple park14. Arctic iceberg bay15. Rainy city street16. Seaside lighthouse cliff17. Wheat farm at dawn18. Busy fishing harbor19. Rooftop greenhouse garden20. Old stone bridge21. Hidden jungle ruins22. Canyon river bend23. Tulip windmill village24. Glacier lake reflection25. Safari elephant herd26. Panda bamboo grove27. Penguin ice colony28. Butterfly meadow path29. Owl in moonlight30. Farm red barn31. Snow-covered pine forest32. Cliffside coastal road33. Starry desert campsite34. Cherry blossom avenue35. Ancient library hall36. Space station window view37. Underwater shipwreck scene38. Carnival ferris wheel39. City skyline sunrise40. Rustic kitchen table41. Pottery studio shelves42. Mountain goat ridge43. Camel caravan dunes44. Vineyard hillside rows45. Ice cream parlor46. Pirate ship deck47. Suburban street in spring48. Riverside picnic spot49. Foggy marsh boardwalk50. Volcano crater rim**Style options (50)**1. Photorealistic cinematic lighting2. Vintage film grain3. Noir high-contrast monochrome4. Pastel watercolor wash5. Bold pop art6. Retro 80s synthwave7. Minimalist flat illustration8. Ukiyo-e woodblock print9. Art nouveau ornamental10. Art deco geometric elegance11. Surreal dreamscape12. Dark fantasy matte painting13. Bright kawaii anime14. Gritty cyberpunk neon15. Steampunk brass and gears16. Low-poly 3D render17. Isometric voxel style18. Claymation stop-motion look19. Pencil sketch crosshatching20. Charcoal drawing texture21. Ink line art22. Gouache poster style23. Oil painting impasto24. Fresco mural aesthetic25. Pixel art 16-bit26. Glitchcore digital distortion27. Holographic iridescent finish28. Paper cut collage29. Stained glass mosaic30. Hand-drawn doodle style31. Children's book illustration32. Botanical scientific illustration33. Blueprint technical drawing34. Architectural visualization clean35. Product studio photography36. Fashion editorial look37. Documentary photojournalism38. Golden hour landscape39. Moonlit night ambience40. Misty atmospheric fog41. Desert dust tones42. Tropical vibrant palette43. Scandinavian soft tones44. Brutalist graphic design45. Bauhaus primary shapes46. Japanese zen minimalism47. Medieval illuminated manuscript48. Rococo ornate detail49. Futuristic concept art50. Hyperreal macro photography### 2. Generate the imageRun the script with your prompt and optional output folder. If no output folder is provided, the image is saved under `images`.```bashnode scripts/generate_puzzle_image.js --prompt "YOUR_PROMPT_HERE" --out "optional/output/folder" --name "my-image.png"```Use this default numbering logic for `--name` argument:1. Scan `images/generated-*.png`.2. Extract numeric suffixes from matching filenames.3. Compute the next `N` as `max + 1` (default to `1` if no matches exist).4. Generate with filename `images/generated-{N}.png`.### 3. Post-image-generation ruleAfter generating images, derive `N` from current generated files before updating code:1. Scan `images/generated-*.png`.2. Extract numeric suffixes and set `N` to the maximum suffix (example: latest is `generated-14.png` => `N = 14`).3. Replace hard-coded magic numbers with `N` in the `index.html` file, for example: ```js const GENERATED_NUMBERS = 15; // Replace 15 with N ```## Script Notes- Requires `GEMINI_API_KEY` in the environment.- Expects the Gemini response to include `inlineData.data` (base64 image).- Saves a `.png` by default and derives the extension from `inlineData.mimeType` when available.

doggy8088/memory-puzzle/tree/main/.agents/skills/puzzle-image-generator commit 3821a00d8a

Frequently asked questions

npx skillmds@latest add doggy8088/puzzle-image-generator