Selling emoji and sticker packs on KakaoTalk, LINE, Telegram, and OGQ is a real business — creators earn thousands monthly from packs that resonate. But the pipeline from idea to store-ready ZIP is brutal: character design, 45+ sticker variations, background removal, platform-specific resizing, localized metadata, and separate bundles for every marketplace. awesome-emoji collapses that entire workflow into a single CLI command powered by Gemini AI.

:sparkles: Features

:brain: AI Expert Panel

Three specialized AI personas — Market Analyst, Art Director, and Cultural Expert — independently evaluate your concept before a Chief Creative Director synthesizes their insights into a unified creative strategy. This isn't a single prompt — it's a structured debate.

:art: 5 Visual Styles

Choose from Chibi/Meme, Clean Cartoon, Pixel/Retro, Expressive Anime, or Original/Balanced — or let the AI pick the best fit for your concept and target market.

:package: 6 Platform Targets

Export submission-ready packs for KakaoTalk, LINE, Telegram, and OGQ with correct dimensions, formats, thumbnails, and tab images — all from one command.

:scissors: ML-Free Background Removal

Custom Sobel edge detection + flood-fill algorithm removes backgrounds cleanly with zero ML dependencies. Includes automatic defringing to eliminate color halos around edges.

:globe_with_meridians: AI Metadata with SEO

Auto-generates 3 metadata variants per pack with localized titles, descriptions, and 20 SEO tags in Korean, Japanese, and Traditional Chinese — each self-scored so you can pick the strongest option.

:robot: Machine-Readable Output

Pass --json for structured NDJSON on stdout — every pipeline stage, progress event, and result streamed as machine-readable records. Built for AI agent integration and CI/CD workflows.

:gear: Auto Decision Engine

Intelligent defaults resolve optimal platforms, language, and visual style from your target market. Specify --market kakaotalk and it handles the rest.

:test_tube: Battle-Tested

2,184 lines of tests covering 145+ edge cases including integration tests, service validation, and full pipeline reliability checks.

:computer: Demo

$ awesome-emoji generate -c "a cheerful capybara wearing a tiny hat"

  awesome-emoji-cli v0.1.0

  Concept:   a cheerful capybara wearing a tiny hat
  Style:     auto (AI will decide)
  Platforms: auto (market analysis)

  [1/8] Analyzing concept...
    ├─ Market Analyst:   High demand for animal characters in KR/JP markets
    ├─ Art Director:     Recommending "Clean Cartoon" style for broad appeal
    └─ Cultural Expert:  Capybara trending across all target regions ✓

  [2/8] Generating character sheet...          done (4.2s)
  [3/8] Selecting visual style... Clean Cartoon
  [4/8] Ideating emote expressions...          done (3.1s)
    └─ 24 expressions generated (happy, sad, angry, love, sleeping...)

  [5/8] Generating sticker images...           done (28.6s)
    └─ 24/24 stickers rendered

  [6/8] Post-processing images...              done (6.3s)
    ├─ Background removal (Sobel + flood-fill)
    ├─ Resized for 6 platform targets
    └─ Quality check passed

  [7/8] Generating metadata...                 done (5.1s)
    ├─ 3 title variants per locale (ko, ja, zh-TW)
    ├─ SEO tags generated
    └─ Best variant auto-selected (score: 0.94)

  [8/8] Exporting platform bundles...          done (1.8s)

  ✔ Complete! Generated 6 submission-ready packs:

  output/
  ├── ogq-sticker-740x640/           24 images + metadata.json
  ├── line-sticker-370x320/          24 images + metadata.json
  ├── line-emoji-180x180/            24 images + metadata.json
  ├── kakaotalk-emoticon-360x360/    24 images + metadata.json
  ├── kakaotalk-mini-180x180/        24 images + metadata.json
  └── telegram-static-512x512/       24 images + metadata.json

  Total: 144 images | 6 ZIPs | 3 locales | 49.3s elapsed

:rocket: Quick Start

1. Install

npm install -g awesome-emoji-cli

2. Set your Gemini API key

awesome-emoji config set-key YOUR_GEMINI_API_KEY
# Persists to ~/.awesome-emoji/config.json — you only do this once

3. Generate your first emoji pack

awesome-emoji generate -c "a cheerful orange cat who loves coffee"

4. Find your output

ls ./output/
# main_character.png  stickers/  processed/  metadata.json  session.json
# Plus submission-ready ZIPs for every target platform

:bulb: Usage

Basic — just describe your concept:

awesome-emoji generate -c "a sleepy panda studying for exams"

Target a specific market and language:

awesome-emoji generate \
  -c "cute bread character with expressive reactions" \
  --market kakaotalk \
  --language ko \
  --platforms kakaotalk_emoticon,kakaotalk_mini

Advanced — reference image, custom outline, JSON output:

awesome-emoji generate \
  -c "my existing mascot as a sticker pack" \
  --reference-image ./mascot.png \
  --outline black \
  --outline-thickness 5 \
  --outline-opacity 80 \
  --json \
  -o ./release/v2

Post-process existing images:

awesome-emoji postprocess -s ./my-stickers/ --market kakaotalk

Dry run (no API key needed):

awesome-emoji generate -c "test concept" --dry-run

:earth_asia: Platforms

| Platform | Target | Dimensions | Format | Sticker Count | |:---:|:---:|:---:|:---:|:---:| | OGQ | Sticker | 740 x 640 | PNG | 16–40 | | LINE | Sticker | 370 x 320 | PNG | 8–40 | | LINE | Emoji | 180 x 180 | PNG | 40 | | KakaoTalk | Emoticon | 360 x 360 | PNG | 16–40 | | KakaoTalk | Mini Emoticon | 180 x 180 | PNG | 16–40 | | Telegram | Static Sticker | 512 x 512 | PNG/WEBP | up to 120 |

• OGQ Sticker — 740x640px, transparent PNG, metadata in Korean
• LINE Sticker — 370x320px (with surrounding margin), transparent PNG
• LINE Emoji — 180x180px, transparent PNG, strict sizing requirements
• KakaoTalk Emoticon — 360x360px, transparent PNG, Korean metadata required
• KakaoTalk Mini — 180x180px, same as above but smaller format
• Telegram Static — 512x512px, PNG or WEBP, one side must be exactly 512px

All exports include correctly structured metadata files ready for direct submission to each platform's creator portal.

:building_construction: Architecture

Your text concept enters the Expert Panel, where three Gemini-powered AI personas — a Market Analyst, Art Director, and Cultural Expert — independently evaluate your idea against target market dynamics, then a Chief Creative Director synthesizes their recommendations into a unified strategy with a chosen visual style. That strategy drives the Generation Pipeline: character creation, 45+ emote variations rendered in one of five visual styles (Chibi/Meme, Clean Cartoon, Pixel/Retro, Expressive Anime, or Original/Balanced), each processed through Sharp-based background removal (Sobel edge detection + flood-fill, no ML) and configurable outline generation. The pipeline then generates three self-scored metadata variants with localized titles, descriptions, and SEO tags. Finally, the Export Bundler resizes every sticker to each platform's exact spec, packages thumbnails and tab images, and writes submission-ready ZIPs — one per platform format.

flowchart LR
    A["fa:fa-keyboard Text Concept"] --> B["1. Concept\nAnalysis"]
    B --> C["2. Character\nGeneration"]
    C --> D["3. Style\nSelection"]
    D --> E["4. Emote\nIdeation"]
    E --> F["5. Sticker\nGeneration"]
    F --> G["6. Post-Processing"]
    G --> H["7. Metadata\nGeneration"]
    H --> I["8. Platform\nExport"]

    I --> J["fa:fa-file-zipper OGQ\n740x640"]
    I --> K["fa:fa-file-zipper LINE Sticker\n370x320"]
    I --> L["fa:fa-file-zipper LINE Emoji\n180x180"]
    I --> M["fa:fa-file-zipper KakaoTalk\n360x360"]
    I --> N["fa:fa-file-zipper KakaoTalk Mini\n180x180"]
    I --> O["fa:fa-file-zipper Telegram\n512x512"]

    subgraph ai ["Gemini AI"]
        B
        C
        D
        E
        F
        H
    end

    subgraph processing ["Sharp Pipeline"]
        G
    end

    style ai fill:#E8D5F5,stroke:#8E75B2,stroke-width:2px
    style processing fill:#E8F5E9,stroke:#4CAF50,stroke-width:2px
    style A fill:#FFF3E0,stroke:#FF9800,stroke-width:2px
    style I fill:#E3F2FD,stroke:#2196F3,stroke-width:2px

:wrench: Configuration

awesome-emoji resolves your Gemini API key from three sources, checked in order:

1. CLI flag — --api-key YOUR_KEY (highest priority, useful for CI)
2. Environment variable — GEMINI_API_KEY
3. Config file — ~/.awesome-emoji/config.json (set once with awesome-emoji config set-key)

# Save your key persistently
awesome-emoji config set-key YOUR_KEY

# Check which key is active (masked output)
awesome-emoji config get-key

# Remove a stored key
awesome-emoji config delete-key

Model Fallback: The CLI targets gemini-3.1-pro-preview for text reasoning and gemini-2.5-flash-image for image generation. If a primary model is unavailable or rate-limited, the fallback chain automatically retries with gemini-2.5-flash — no configuration needed.

:test_tube: Testing

npm test

The test suite includes 145+ test cases across three categories:

• CLI tests — End-to-end subprocess tests validating command parsing and output
• Core tests — Direct module tests for market resolver, auto-decision engine, and constants
• Edge case tests — Comprehensive service validation covering boundary conditions and error paths

:handshake: Contributing

Contributions are welcome and appreciated. Whether it's a bug fix, a new platform format, or a better background removal algorithm — open an issue or submit a pull request. All skill levels belong here.

:scroll: License

Released under the MIT License. Use it, modify it, ship it.