What is GAB?
GAB (Generic AALang Builder) is an intelligent agent compiler that helps you build AALang-based products including tools, games, agents, protocols, and more. GAB uses a structured 4-mode workflow to guide you through the entire development process, from initial concept to final product compiled to AALang.
What Can You Build with GAB?
AALang is a general-purpose programming language - you can build virtually anything that can be expressed computationally. GAB helps you create:
- 🎮 Games - Interactive games powered by LLM agents
- 🛠️ Tools - Utilities and applications that leverage LLM capabilities
- 🤖 Agents - Custom LLM agents with specific behaviors and modes
- đź“‹ Protocols - Communication and interaction protocols
- đź’¬ Communication Patterns - Patterns for agent-to-agent or agent-to-user communication
- 📦 Any AALang-based Product - Anything that conforms to AALang specifications
Building Specific Product Types
Creating Games
GAB excels at creating interactive games. Simply describe your game concept, and GAB will:
- Design the game modes and states
- Create actor personas for game logic
- Implement user interaction patterns
- Generate the complete game specification
Example: “Create a trivia game with multiple categories and difficulty levels”
Creating Tools
Build tools that leverage LLM capabilities:
- Describe the tool’s purpose
- Specify input/output requirements
- Define the workflow
- GAB generates the tool specification
Example: “Create a code review tool that analyzes code and provides suggestions”
Creating Agents
Design custom LLM agents for specific tasks:
- Define the agent’s purpose
- Specify modes and behaviors
- Design persona interactions
- Generate the agent specification
Example: “Create a customer support agent that handles common questions”
Executing AALang Code
Once GAB generates your AALang specification:
- Load the Generated File: Load the
.jsonldfile into your LLM - Execute: The LLM interprets and executes the AALang code
- Interact: Use the product as designed
AALang specifications are MCP and A2A ready, meaning they integrate seamlessly with Model Context Protocol tooling for enhanced LLM interactions and support native Agent-to-Agent communication via gossip-based P2P protocols for distributed execution.