Skip to content

Antigravity Skills Engine (ASE) - The ultimate Claude Skills library for AI agents featuring 16+ industrial-grade professional tracks.

License

Notifications You must be signed in to change notification settings

VidyFoo/antigravity-skill-engine

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

⚡ Antigravity Skills Engine (ASE)

Antigravity Skills Engine (ASE) is a high-performance AI Agent Skills Library and Workflow Engine. It transforms a single LLM into an industrial-grade Expert System through a dynamic registry of specialized skills. It acts as the Central Nervous System for your AI Agent.

Version: 3.2.0 | Status: Production Ready | License: MIT


🏛️ Core Architecture: Registry V3

ASE 3.x introduces a Category-based Registry Architecture that organizes 15+ skills into 5 strategic domains. The /ase hyper-workflow dynamically reads this registry to discover, identify, and activate the appropriate expert for any task.

🗂️ The 5 Domains of Expertise

Code Domain Focus Skills Count
1-dev 🔧 DEV (Development) Feature architecture, fullstack guide, session lifecycle 3
2-audit 🔍 AUDIT (Quality) Code auditing, scientific debugging, webapp testing 3
3-sys ⚙️ SYS (System) Ecosystem management, skills management, doc bootstrapping, MCP building 4
4-tools 🛠️ TOOLS (Utilities) Doc writing, visual designing, PPTX creating 3
external 🌐 EXTERNAL (Dependencies) Brand guidelines, internal comms, meeting insights 3

📋 Full Skills Registry

1-dev: Development Domain
ID Skill Description
1.1 architecting Architects new features through the full lifecycle (design → plan → implement)
1.2 fullstack-guide Fullstack development encyclopedia (React/TS/Node/Supabase)
1.3 session-managing Auto-triggered session lifecycle management (INHERIT + DEPOSIT)
2-audit: Audit Domain
ID Skill Description
2.1 code-auditing Systematic auditing & code optimization (Code Review / Simplifier / Council)
2.2 scientific-debugging Scientific debugging: Observe → Hypothesize → Experiment → Fix
2.3 webapp-testing Playwright-based toolkit for testing local web applications
3-sys: System Domain
ID Skill Description
3.1 ecosystem-managing System ecosystem engineer (Release / Maintain / Purge)
3.2 skills-managing Create and manage ASE skills following best practices
3.3 doc-bootstrapping One-click project documentation system initialization
3.4 mcp-building Guide for creating high-quality MCP servers
4-tools: Tools Domain
ID Skill Description
4.1 doc-writing Technical doc engineering (Diátaxis / PDF / DOCX / XLSX)
4.2 visual-designing Visual design toolbox (Canvas / Themes / Design System / Artifacts)
4.3 pptx-creating PPTX presentation creation, editing & analysis (html2pptx / OOXML)
external: External Dependencies
ID Skill Description
E.1 brand-guidelines Official brand colors and typography
E.2 internal-comms Internal communications templates (status reports, newsletters, etc.)
E.3 meeting-insights-analyzer Behavioral patterns and communication insights from meetings

📂 Project Structure

.agent/
├── skills/
│   ├── registry.json           # Single Source of Truth (V3)
│   ├── 1-dev/                  # Development Domain
│   │   ├── architecting/
│   │   ├── fullstack-guide/
│   │   └── session-managing/
│   ├── 2-audit/                # Audit Domain
│   │   ├── code-auditing/
│   │   ├── scientific-debugging/
│   │   └── webapp-testing/
│   ├── 3-sys/                  # System Domain
│   │   ├── doc-bootstrapping/
│   │   ├── ecosystem-managing/
│   │   ├── mcp-building/
│   │   └── skills-managing/
│   ├── 4-tools/                # Tools Domain
│   │   ├── doc-writing/
│   │   └── visual-designing/
│   └── external/               # External Dependencies
│       ├── brand-guidelines/
│       ├── internal-comms/
│       └── meeting-insights-analyzer/
└── workflows/
    └── ase.md                  # The /ase Hyper-Workflow (Dynamic Router)

🛠️ Usage

Invoke the Central Nervous System: /ase

Simply type /ase in your chat. The engine will enter Discovery Mode, scanning registry.json and presenting you with the Expert Matrix.

Then, ask naturally:

"Review my code for error handling bugs." → Activates code-auditing → Code Review Mode

"Design a premium glassmorphism landing page." → Activates visual-designing → Premium Mode

"Debug this intermittent crash." → Activates scientific-debugging → Hypothesis-driven workflow

Lifecycle Phases

The ASE operates through a structured lifecycle:

Phase Name Purpose
-1 INHERIT Load project context (Rule1, Memory)
0 IDENTIFY Match user intent to skill via registry
1 EXECUTE Activate skill protocol with resources
F DEPOSIT Synthesize learnings, harvest reusable artifacts

💎 Key Features (V3.1)

  • Category-based Organization: Skills are grouped into logical domains with numerical prefixes for easy navigation
  • Protocol-First Design: Skills enforce strict execution protocols with resource loading
  • Progressive Disclosure: Heavy knowledge is offloaded to resources/ files until needed
  • Auto-triggered Skills: Certain skills (e.g., session-managing) activate automatically
  • Trace Mode: Clear visibility into ASE execution with phase markers

🚀 Creating New Skills

Want to add a new expert?

  1. Summon the Skills Manager: Invoke skill 3.2
  2. It will generate the standard SKILL.md and register it in registry.json automatically
  3. The new skill is instantly available via /ase

📚 Documentation


Antigravity Lab © 2026 | Speeding up your workflow with precision.

About

Antigravity Skills Engine (ASE) - The ultimate Claude Skills library for AI agents featuring 16+ industrial-grade professional tracks.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published