Skip to content

Latest commit

 

History

13 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 

Repository files navigation

App Icon & Image Skills

English | 简体中文

skills.sh

Reusable creative-production skills centered on a modular app-icon system, with supporting workflows for transparent raster assets, code-ready ANSI terminal art, and painterly full-color glyph mosaics.

Available skills

Skill Purpose Install
app-icon-design Orchestrate one app-icon identity across optional themes, visual styles, Apple, Android, Web/PWA, Windows, and other targets. npx skills add fine405/skills --skill app-icon-design
app-icon-theme Create coherent theme-driven variants of an existing app icon while preserving its identity and keeping themes independent of style and platform. npx skills add fine405/skills --skill app-icon-theme
imagegen-transparent Generate or extract clean transparent PNG/WebP assets with chroma-key removal, soft alpha matting, despill, and validation. npx skills add fine405/skills --skill imagegen-transparent
imagegen-ansi Convert references or generated raster art into transparent ANSI-style assets, terminal half-block output, and executable JavaScript previews. npx skills add fine405/skills --skill imagegen-ansi
imagegen-glyph-mosaic Analyze visual references and generate full-color glyph-mosaic illustrations with a coherent character grid, directional glyph logic, and adjustable palette roles. npx skills add fine405/skills --skill imagegen-glyph-mosaic

App Icon System

The app-icon skills separate stable product identity from semantic themes, visual styles, and platform production rules:

identity → optional theme-* → optional style-* → platform-* → final QA

App Icon Design

Use app-icon-design as the orchestrator. Build one recognizable product identity, then adapt it to each target without mixing platform specifications into theme or style rules.

Workflow

  1. Reduce the app to an audience, core job, promise, category cue, and differentiator.
  2. Lock an identity invariant: metaphor, silhouette, proportions, brand-color role, and one signature detail.
  3. Load only the requested platform modules for Apple, Android, Web/PWA, or Windows.
  4. Resolve an optional semantic theme with app-icon-theme, then apply an installed style module or task-local style brief.
  5. Generate or edit the artwork, test target masks and small sizes, and report remaining packaging steps honestly.

Bundled styles include soft neumorphism, discomorphism, and Jelly 3D. New visual directions remain independent style-*.md modules.

App Icon Theme

Use app-icon-theme to reinterpret a user-owned icon through a music genre, season, mood, event, sport, profession, or content category without losing the source identity.

Workflow

  1. Lock the source silhouette, internal cutouts, proportions, layout, and brand-color roles.
  2. Select one primary semantic cue and a transformation mode: material, structural analogy, or controlled accessory.
  3. Set subtle, balanced, or expressive intensity; balanced is the default.
  4. Apply an independently selected rendering style, then check identity and theme legibility at small sizes.

The first bundled theme family is theme-music-genre.md. Future theme families use independent theme-*.md modules.

ImageGen Transparent

Use Codex ImageGen or a compatible image generator to create a controlled chroma-key source, then remove the background locally and validate the result.

Workflow

  1. Choose a key color absent from the subject, normally #00ff00.
  2. Generate the subject on an exact, flat key background with no shadows or reflections.
  3. Sample the actual border color and build a soft alpha matte.
  4. Remove key-color spill and optionally contract the edge by 1 px.
  5. Validate RGBA mode, transparent corners, subject coverage, clipping, and edge quality.

Snowflake sprite-sheet example

The example recreates four supplied snowflake forms, adds twelve common snow-crystal forms, and produces a ranked 4×4 transparent sprite sheet.

Generated chroma-key source Validated transparent output
Blue snowflake sprite sheet on a green chroma-key background Transparent 4 by 4 blue snowflake sprite sheet
1254×1254 RGB source 1256×1256 RGBA output with 314×314 cells

ImageGen ANSI

Use direct alpha conversion when a source already has a clean silhouette. Use ImageGen when the request needs a deliberate terminal-cell reinterpretation, then derive both the raster asset and executable terminal renderer from the same sampled bitmap.

Workflow

  1. Inspect the reference and preserve its silhouette, proportions, spacing, and exact text when present.
  2. Generate white terminal-cell geometry on a flat removable key background.
  3. Extract and validate a transparent RGBA raster.
  4. Sample its alpha channel and pack two vertical pixels into , , and .
  5. Preview directly in the terminal or emit an executable .mjs module.

Mountain-and-sun example

This example uses an original, generic mountain-and-rising-sun badge with no text or brand identity.

  1. Generate a flat reference badge.
  2. Reinterpret it as coarse white ANSI cell geometry on #00ff00.
  3. Sample the generated border (#03ed0b), remove it, and validate four transparent corners.
  4. Convert the alpha mask to a 48×50 bitmap and 25 terminal rows.
  5. Emit both plain ANSI text and an adaptive JavaScript renderer.

Prompt summary:

Preserve the generic circular badge, two mountain peaks, and rising sun while translating all curves into deliberate terminal-cell steps. Render pure white on a perfectly flat green key background with no text, brand identity, shadows, gradients, or decoration.

Original reference Generated chroma-key ANSI art Validated transparent ANSI art
Generic mountain and rising sun badge reference White ANSI mountain and sun badge on a green chroma-key background Transparent white ANSI mountain and sun badge

Run the code-generated terminal preview:

node ./skills/imagegen-ansi/examples/mountain-sun-ansi.mjs

The corresponding plain-text output is available at mountain-sun-ansi.txt.

ImageGen Glyph Mosaic

Use style references to separate stable visual DNA from adjustable subject and palette variables, then generate painterly raster scenes whose image-forming marks remain visible monospaced glyphs.

Workflow

  1. Label supplied images as style references or edit targets.
  2. Extract the stable grid, dual-scale readability, directional glyph logic, restrained palette, and print surface.
  3. Keep subject, composition, weather, and palette roles adjustable.
  4. Generate each distinct scene with its own prompt while reusing the same style core.
  5. Validate the result both as a small coherent scene and as a close-up character grid.

Polychrome glyph-mosaic examples

All three examples use the same style core with scene-specific composition and palette-role blocks. They are original outputs generated for this skill.

Forest Snow night Golden mountain
Primeval forest rendered as a full-color glyph mosaic Moonlit snow valley and cabin rendered as a full-color glyph mosaic Golden sunrise mountain rendered as a full-color glyph mosaic
Moss, dusty teal, ochre, and ivory Indigo, icy blue-gray, cream, and restrained amber Powder blue, glacier ivory, warm gold, and deep umber

The reusable template is in references/prompt-template.md, and the example scene prompts are in examples/prompts.md.

Compatibility and requirements

  • Codex's built-in ImageGen is the default generator; another authorized generator may be used when it preserves the active skill's prompt constraints, reference roles, and local output workflow.
  • Chroma-key extraction and ANSI conversion require Python 3 and Pillow; glyph-mosaic generation has no additional local runtime dependency.
  • App-icon concept work has no additional local runtime dependency; production handoff requires checking the current official requirements for every target platform.
  • Each skill includes agents/openai.yaml with Codex-facing display metadata.

Repository structure

skills/
├── app-icon-design/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/
│       ├── platform-android.md
│       ├── platform-apple.md
│       ├── platform-web.md
│       ├── platform-windows.md
│       ├── module-style-template.md
│       ├── style-discomorphism.md
│       ├── style-jelly-3d.md
│       └── style-soft-neumorphic.md
├── app-icon-theme/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/
│       ├── module-theme-template.md
│       └── theme-music-genre.md
├── imagegen-ansi/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   ├── scripts/
│   │   ├── raster_to_ansi.py
│   │   └── remove_chroma_key.py
│   └── examples/
│       ├── mountain-sun-reference.png
│       ├── mountain-sun-ansi-chroma.png
│       ├── mountain-sun-ansi.png
│       ├── mountain-sun-ansi.txt
│       └── mountain-sun-ansi.mjs
├── imagegen-glyph-mosaic/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   ├── references/prompt-template.md
│   └── examples/
│       ├── prompts.md
│       ├── forest.png
│       ├── snow-night.png
│       └── golden-mountain.png
└── imagegen-transparent/
    ├── SKILL.md
    ├── agents/openai.yaml
    ├── scripts/remove_chroma_key.py
    └── examples/
        ├── snowflake-sprite-sheet-chroma.png
        └── snowflake-sprite-sheet.png

License

MIT

About

Reusable image-generation skills for coding agents.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages