Pixel Asset Master logo

Pixel Asset Master

Community
424431185
pixel-asset-master

AI-driven 2D pixel game asset generation system. Converts game design descriptions into pixel art sprites, tilesets, animations, and UI elements through multi-role collaboration. Exports as sprite sheets and individual PNGs. Use when user asks to "create pixel art", "generate game sprites", "make tileset", "生成像素素材", "做游戏瓦片", "制作精灵图", or mentions "pixel-asset-master".

Overview

Publisher424431185
Repositorypixel-asset-master-skills
Skill namepixel-asset-master
Stars
102
Forks
14
Bundled files
16
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 16 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by 424431185 on GitHub. Read the source before you install it.

Installation

Install the Pixel Asset Master AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/424431185/pixel-asset-master-skills/tree/main/skills/pixel-asset-master
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/424431185/pixel-asset-master-skills.git /tmp/pixel-asset-master-skills
mkdir -p .claude/skills
cp -r /tmp/pixel-asset-master-skills/skills/pixel-asset-master .claude/skills/pixel-asset-master
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Pixel Asset Master in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Pixel Asset Master on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Pixel Asset Master is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Pixel Asset Master Skill

AI-driven 2D pixel game asset generation system. Converts game design descriptions into pixel art sprites, tilesets, animations, and UI elements through multi-role collaboration and exports as sprite sheets / individual PNGs.

Core Pipeline: Game Description → Create Project → Style Confirmation → Strategist → [Image_Generator] → Executor → Post-processing → Export

[!CAUTION]

🚨 Global Execution Discipline (MANDATORY)

This workflow is a strict serial pipeline. The following rules have the highest priority — violating any one of them constitutes execution failure:

  1. SERIAL EXECUTION — Steps MUST be executed in order; the output of each step is the input for the next. Non-BLOCKING adjacent steps may proceed continuously once prerequisites are met
  2. BLOCKING = HARD STOP — Steps marked ⛔ BLOCKING require a full stop; the AI MUST wait for an explicit user response before proceeding
  3. NO CROSS-PHASE BUNDLING — Cross-phase bundling is FORBIDDEN
  4. GATE BEFORE ENTRY — Each Step has prerequisites (🚧 GATE) listed at the top; these MUST be verified before starting that Step
  5. NO SPECULATIVE EXECUTION — "Pre-preparing" content for subsequent Steps is FORBIDDEN
  6. NO SUB-AGENT GENERATION — Executor Step 6 pixel art generation is context-dependent and MUST be completed by the current main agent end-to-end
  7. SEQUENTIAL ASSET GENERATION — Assets MUST be generated sequentially in one continuous pass
  8. SPEC_LOCK RE-READ PER ASSET — Before generating each asset, Executor MUST re-read spec_lock.md. All colors / sizes / palettes MUST come from this file

[!IMPORTANT]

🌐 Language & Communication Rule

  • Response language: match the user's input. Explicit user override takes precedence.
  • Template format: design_spec.md MUST follow its original English template structure regardless of conversation language. Content values may be in the user's language.

[!IMPORTANT]

🔌 Compatibility With Generic Coding Skills

  • pixel-asset-master is a repository-specific workflow, not a general application scaffold
  • Do NOT create .worktrees/, tests/, branch workflows, or generic engineering structure by default
  • On conflict with a generic coding skill, follow this skill unless the user explicitly says otherwise

Main Pipeline Scripts

ScriptPurpose
${SKILL_DIR}/scripts/project_manager.pyProject init / validate / manage
${SKILL_DIR}/scripts/palette_analyzer.pyPalette extraction and validation
${SKILL_DIR}/scripts/asset_validator.pyPixel art asset quality check
${SKILL_DIR}/scripts/sprite_sheet.pySprite sheet packing and export
${SKILL_DIR}/scripts/image_gen.pyAI image generation (multi-provider)
${SKILL_DIR}/scripts/finalize_assets.pyAsset post-processing (quantize, clean, index)

For complete tool documentation, see ${SKILL_DIR}/scripts/README.md.

Template Index

IndexPathPurpose
Palette library${SKILL_DIR}/templates/palettes/palettes_index.jsonQuery available pixel art color palettes
Size presets${SKILL_DIR}/templates/sizes/sizes_index.jsonQuery standard pixel art dimensions
Sprite templates${SKILL_DIR}/templates/sprites/sprites_index.jsonQuery sprite layout templates (character, tile, item, UI)

Standalone Workflows

WorkflowPathPurpose
create-paletteworkflows/create-palette.mdStandalone palette creation workflow
batch-animateworkflows/batch-animate.mdBatch animation frame generation

Pipeline Steps

Step 1: Source Content Processing

🚧 GATE: User provides game design description (text/image/reference)

Input TypeProcessing
Text descriptionDirect use as design brief
Reference imagesAnalyze style, palette, size via palette_analyzer.py
Game design docExtract asset requirements list
Existing spritesAnalyze and extend existing style

Output: Structured asset requirements list


Step 2: Project Initialization

🚧 GATE: Step 1 complete

bash
python ${SKILL_DIR}/scripts/project_manager.py init <project_name> --size 32x32 --palette default

Project structure:

projects/<name>_<size>_<date>/
├── design_spec.md          # Human-readable design narrative
├── spec_lock.md            # Machine-readable execution contract
├── images/                 # User-provided reference images
├── assets/                 # Generated pixel art assets (PNG)
│   ├── characters/
│   ├── tiles/
│   ├── items/
│   ├── ui/
│   ├── effects/
│   └── backgrounds/
├── animations/             # Animation frame sequences
├── sheets/                 # Packed sprite sheets
├── notes/                  # Design notes per asset
└── exports/                # Final export directory

If user provides reference images:

bash
python ${SKILL_DIR}/scripts/project_manager.py import-sources <project_path> <files...> --move

Step 3: Style Confirmation

🚧 GATE: Step 2 complete

Select pixel art style direction:

StyleDescriptionTypical SizePalette
NES Classic8-bit console style, hard edges16x16 / 32x324-8 colors per sprite
SNES Retro16-bit console, more detail32x32 / 64x648-16 colors per sprite
Modern PixelClean, contemporary indie32x32 / 64x6416-32 colors per sprite
MinimalistUltra-simple, few pixels8x8 / 16x162-4 colors per sprite
Dense DetailRich detail, large sprites64x64 / 128x12824-48 colors per sprite
High-Res PixelUltra-detailed, illustration-grade256x256 / 512x51264-96 colors per sprite
Cinematic PixelPrint/billboard quality1024x1024 / 2048x2048128-256 colors per sprite

⛔ BLOCKING — Present style options to user, wait for confirmation.


Step 4: Strategist — Six Confirmations

🚧 GATE: Step 3 confirmed

The Strategist formulates Six Confirmations for the pixel art project:

#ConfirmationDescription
1Canvas SizeBase sprite size per frame. Available presets: 8x8 / 16x16 / 24x24 / 32x32 / 48x48 / 64x64 / 128x128 / 256x256 / 512x512 / 1024x1024 / 2048x2048. Custom sizes allowed (max 2048x2048). For character sprite sheets, prefer ≥ 64x64 for sufficient detail.
2Asset CountTotal number of assets to generate
3Target PlatformGame engine / runtime (Unity, Godot, RPG Maker, web, etc.)
4Color PaletteSpecific palette or auto-derived from references
5Art StylePixel art sub-style (outlined / outlineless / dithered / cel-shaded)
6Animation NeedsFor each character asset, confirm the following:

Animation detail breakdown (for character assets):

ItemOptionsNotes
Facing direction4-dir (up/down/left/right) / 8-dir / side-only (left+right) / single (front-facing only)Determines how many directional rows the sprite sheet needs
Action listSelect from common actions below, or specify customEach action = one row in the sprite sheet
Frames per actionTypical 4-8 frames per action; idle can be 2-4, run 6-8, attack 4-6More frames = smoother but larger sheet

Common character actions:

ActionTypical FramesDescription
idle2-4Standing breathing / subtle motion
walk4-6Slow-paced walking cycle
run6-8Fast running cycle
jump4-6Jump ascent + peak
fall2-4Falling / descent
attack4-6Melee / ranged attack
hurt2-3Taking damage reaction
die4-6Death animation
cast4-6Magic / skill casting
interact2-4Opening chest / pushing / pulling
climb4-6Ladder / wall climbing
crouch2-3Crouching / sneaking
roll4-6Dodge roll
swim4-6Swimming cycle

⛔ BLOCKING — Present Six Confirmations to user, wait for explicit confirmation.

After confirmation, generate:

  • design_spec.md — Human-readable design narrative
  • spec_lock.md — Machine-readable execution contract

Step 5: Image Acquisition

🚧 GATE: Step 4 confirmed

ScenarioAction
User provided referencesAlready in images/, skip
Need additional referencesUse image_gen.py or web search
No references neededGenerate from description only

Step 6: Executor — Generate Pixel Art Assets

🚧 GATE: Step 5 complete, spec_lock.md exists

Role-specific rules: see references/executor-pixel.md

Generation phases:

  1. Visual Construction Phase: Generate all pixel art PNG assets sequentially
  2. Quality Check Gate: Run asset_validator.py on all assets
  3. Logic Construction Phase: Generate design notes per asset

Per-asset generation:

  • Re-read spec_lock.md before each asset
  • Generate at declared canvas size
  • Apply declared palette (quantize if needed)
  • Save to appropriate assets/ subdirectory

Asset categories:

CategorySubdirectoryTypical Assets
Charactersassets/characters/Player, NPCs, enemies, bosses
Tilesassets/tiles/Ground, walls, decorations, transitions
Itemsassets/items/Weapons, potions, keys, collectibles
UIassets/ui/Buttons, frames, icons, bars, panels
Effectsassets/effects/Explosions, particles, magic, weather
Backgroundsassets/backgrounds/Parallax layers, sky, terrain

🧱 Tile Generation Standards (MANDATORY)

Tiles are the most common asset type and easiest to make look "fake". The following 9 rules must ALL be satisfied for every tile.

1. Canvas & Format
  • Default size: 64×64 RGB (no transparency for ground tiles)
  • Palette: Strict quantization to declared palette in spec_lock.md
  • Anti-pattern: Anti-aliasing, semi-transparent edges, gradients with > palette colors
2. NO Obvious Geometric Patterns ⛔

The #1 cause of "fake-looking" tiles. Forbidden in repeatable terrain:

❌ Anti-pattern✅ Replace with
Centered cross / star / plusRandom off-center features
4 equal quadrantsVoronoi 8-15 irregular cells
Regular sine wavesLow-frequency value noise
Horizontal stripe layersDiagonal/random color blobs
Symmetric corner motifsSingle asymmetric accent
Straight diagonal cracksRandom walk segments

Decorative tiles (e.g., tile_brick) may use geometric patterns, but only when explicitly designed as architecture, not as terrain.

3. Multi-Tier Color Distribution

Each pixel sampled by probability bucket:

  • 70% main color (palette base)
  • 15% dark accent (-1 palette step)
  • 10% light accent (+1 palette step)
  • 5% emphasis color (rare highlight/feature)
4. Low-Frequency Value Noise (Required)

For natural color variation, use value noise + cosine interpolation:

python
nodes = [[rng.uniform(-1, 1) for _ in range(gw)] for _ in range(gh)]
# CRITICAL: wrap edges for seamless tiling
for row in nodes: row[gw-1] = row[0]
nodes[gh-1] = list(nodes[0])
# Then cosine-lerp at each pixel

Recommended scales:

  • Main blob: 8~16 px per node (large irregular regions)
  • Detail: 3~6 px per node (subtle variation)
  • Combine 2 octaves for richer texture
5. Seamless Tiling (4-Edge Wrap) ⛔

Tile must repeat without visible seams. Two enforcement methods:

  • Modulo coordinates: img.putpixel((x % W, y % H), color)
  • Wrap noise nodes: last row/col equals first
  • Test: Place 2×2 instances side-by-side; no visible borders
6. Sparse Feature Points (≤ 1% of pixels)

Cracks, mineral veins, moss spots, water droplets, etc:

  • Random-walk paths for cracks (short, intermittent, never straight)
  • Scattered points (not clustered, not aligned)
  • Total feature pixels ≤ 1% of canvas area
  • Forbidden: Continuous straight lines crossing the tile
7. Low-Noise Visual Coherence

Terrain tiles must feel calm and continuous when shown full-screen:

  • Noise rule: Avoid salt-and-pepper noise, snow-like white speckles, or evenly scattered bright dots.
  • Highlight rule: White / near-white pixels are forbidden for ground and underground tiles unless explicitly requested; use warm ochre, muted gray, or one-step lighter palette colors instead.
  • Continuity rule: Prefer broad, low-frequency color regions and short soft bands over per-pixel randomness.
  • Pixel style rule: Keep hard pixel edges and readable clusters; do not blur, smear, or use anti-aliased gradients.
  • Ink-gongbi rule: For this project, terrain should stay in restrained ink-and-gongbi tones: muted earth, sand, jade, water, gray ink; no neon colors or high-contrast glitter.
8. Edge Color Continuity

Edges within ±1 palette step of interior. Avoid:

  • ❌ Black border line around the tile (creates grid effect when tiled)
  • ❌ Drastically different color in 1-pixel border
  • ✅ Same color distribution on edges as interior
9. Mandatory Validation

Before declaring done, verify:

  • Place 4 copies in 2×2 grid → no seams visible
  • Compare with adjacent tiles (e.g., grass + dirt) → color transition feels natural
  • No symmetric features (rotate 90° / flip → looks different)
  • Quantized to exact palette (palette_analyzer.py)
  • Feature density ≤ 1%
  • No visible white speckle noise when viewed at game fullscreen scale
Reference Implementation

See e:/神谱拾遗/projects/ShenPuShiYi_Scenes_64x64_20260502/gen_tiles_v2.py for working examples of:

  • Voronoi stone path with wrap-distance metric
  • Two-octave noise jade with cosine interpolation
  • Random-walk cracks for clay/bedrock
  • Probability-bucketed multi-tier color distribution

Step 7: Post-processing & Export

🚧 GATE: Step 6 complete, all assets pass quality check

bash
# 1. Post-process assets (quantize, clean borders, index colors)
python ${SKILL_DIR}/scripts/finalize_assets.py <project_path>

# 2. Pack sprite sheets
python ${SKILL_DIR}/scripts/sprite_sheet.py <project_path>

# 3. Export final package
# Output:
#   exports/<project_name>_<timestamp>.zip
#   ├── sprites/           ← Individual PNGs
#   ├── sheets/            ← Packed sprite sheets
#   ├── palettes/          ← Palette files (GPL/PAL/ACT)
#   └── manifest.json      ← Asset manifest with metadata

Export formats:

FormatDescription
Individual PNGsOne file per asset/animation frame
Sprite sheetsPacked grid with manifest
Palette filesGPL / PAL / ACT format
Asset manifestJSON with size, palette, animation metadata

Role Switching Rules

PhaseActive RoleReference File
Step 1-3CoordinatorThis file
Step 4Strategistreferences/strategist.md
Step 5Image_Generator (optional)references/image-generator.md
Step 6Executorreferences/executor-pixel.md
Step 7Post-processorThis file + references/shared-standards.md

Switching protocol: Announce role switch with [Role Switch: <Role>] before starting the phase.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Pixel Asset Master AI skill do?

AI-driven 2D pixel game asset generation system. Converts game design descriptions into pixel art sprites, tilesets, animations, and UI elements through multi-role collaboration. Exports as sprite sheets and individual PNGs. Use when user asks to "create pixel art", "generate game sprites", "make tileset", "生成像素素材", "做游戏瓦片", "制作精灵图", or mentions "pixel-asset-master".

Why use Pixel Asset Master on TypingMind?

Because you install it once and use it with any model. Pixel Asset Master is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Pixel Asset Master in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/424431185/pixel-asset-master-skills/tree/main/skills/pixel-asset-master. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Pixel Asset Master?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Pixel Asset Master?

As many as you like. As long as a model supports skills, you can use Pixel Asset Master with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Pixel Asset Master AI skill free?

Yes. It is published on GitHub by 424431185 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇