Summary

MoAI-ADK addresses traditional AI development challenges like unclear requirements and drifting documentation by providing an AI-powered SPEC-First TDD framework. It automates the entire software lifecycle, from requirements to synchronized, tested documentation, enabling developers to build trustworthy, maintainable software with AI assistance.

How It Works

The framework integrates SPEC-First development using EARS-format specifications with an orchestrated TDD (RED→GREEN→REFACTOR) workflow. It leverages Alfred SuperAgent, featuring 19 specialized AI agents and over 125 enterprise skills, to automate tasks. Key advantages include guaranteed 85%+ test coverage, living documentation, persistent project context, and adherence to TRUST 5 principles, ensuring complete traceability. Agent delegation optimizes large project handling by managing token efficiency.

Quick Start & Requirements

Installation is via uv tool install moai-adk. Prerequisites include Git and GitHub CLI. Project initialization involves moai-adk init <project-name> followed by /alfred:0-project for configuration. Detailed documentation is available at adk.mo.ai.kr.

Highlighted Details

• Features 19 specialized AI agents and 125+ enterprise skills, covering domains from BaaS (11 platforms) and security to DevOps and data analytics.
• Employs agent delegation to drastically reduce LLM token usage (80-85% savings) and accelerate processing for large codebases.
• Enforces SPEC-First development with EARS specifications and automated TDD, guaranteeing 85%+ test coverage.
• Offers 5 adaptive personas (e.g., Alfred for beginners, R2-D2 Partner for pair programming) and supports 25+ languages for user interaction.
• Integrates with modern BaaS platforms and MCP protocols, supporting 8 architectural patterns.

Maintenance & Community

The project is maintained by the MoAI Team. Community support is available via GitHub Issues and GitHub Discussions. A community site is noted as "Coming in November - In Development".

Licensing & Compatibility

Released under the MIT License, permitting commercial and private use with optional attribution.

Limitations & Caveats

Requires the uv tool for installation. While agent delegation solves context token issues, managing extremely large projects might still require careful planning. Some community features are still under development.