Skills AI — AI skill for Claude Code

AI Skills Ecosystem 🧠⚡

**Modular, Token-Efficient Knowledge Base & Skill Ingestion Pipeline for AI Agents** *Ecossistema Modular de Base de Conhecimento e Pipeline de Ingestão de Skills para Agentes de IA*

---

🇺🇸 English

🎯 Why Skills? (Motivation & Core Purpose)

Modern Large Language Models (LLMs) and Autonomous AI Agents (like Antigravity, Claude Code, GitHub Copilot, OpenAI Codex, etc.) are exceptionally capable at reasoning, coding, and problem-solving. However, when faced with large codebases, extensive API documentation, complex RPG rulebooks, or proprietary company manuals, they suffer from two major problems:

1. Context Window Exhaustion & High Token Cost: Dumping entire books or monolithic documentation files into the prompt window wastes thousands of tokens per turn and causes hallucinations or attention degradation ("lost in the middle").
2. Lack of Domain-Specific Structure: Raw, unformatted documents contain noisy layout elements (headers, footers, broken tables, column mixing, fragmented code blocks) that confuse the agent.

The Solution: The Skills Architecture

**Skills** transform raw data (documents, web portals, APIs) into **structured, on-demand micro-knowledge units**:

• Master Index (SKILL.md): Serves as the cognitive map and mental model for the agent.
• On-Demand Chapters (chapters/*.md): Deep technical chapters loaded only when specifically relevant to the current user prompt (~10KB to 40KB each).
• Fast-Path Reference (cheatsheet.md, glossary.md, patterns.md): Instant access to syntaxes, canonical terms, and architectural patterns without reading entire manuals.

---

📦 Installation & Setup

1. Clone the Repository

git clone https://github.com/CharlesCorreaDev/skills-ai.git
cd skills-ai

2. Create Virtual Environment & Install Python Dependencies

# Create and activate virtual environment
python -m venv .venv

# Windows (PowerShell)
.venv\Scripts\Act