Maestro V2 is a streamlined 5-mode orchestration system for Roo Code that enables efficient task delegation and specialized AI assistance through dynamic persona loading.
Maestro V2 brings advanced workflow orchestration to Roo Code through a streamlined 5-mode architecture with dynamic persona loading. The system provides deep expertise without context pollution, making AI assistance more efficient and maintainable.
- 5 Primary Modes: Each optimized for specific tasks
- Dynamic Personas: Load specialized knowledge only when needed
- Zero Context Pollution: Modes stay focused by loading only relevant information
- Workflow Management: Intelligent task routing and state management
- Flexible Architecture: Easy to extend with new personas
- Routes tasks to appropriate specialized modes
- Manages workflow state and context
- Handles user interaction and progress tracking
- No personas needed - pure orchestration
- Analyzes existing codebases and reviews changes
- Two specialized personas:
code-scanner
: Automated analysis and pattern detectioncode-reviewer
: Human-like code review with suggestions
- High-level planning and architecture design
- Technology selection and roadmapping
- Creates comprehensive technical documentation
- No personas - uses broad architectural knowledge
- Simple, straightforward coding tasks
- Rapid prototyping and quick fixes
- No overthinking or complex architectures
- No personas - direct implementation
- Full-featured implementation with specialized knowledge
- Dynamic loading from extensive persona library:
- Frontend (React, Vue, Angular, etc.)
- Backend (Node.js, Python, Java, etc.)
- Database (PostgreSQL, MongoDB, etc.)
- Testing (Jest, Playwright, etc.)
- Debugging (Systematic debugging strategies)
- DevOps (Docker, Kubernetes, etc.)
- Copy the required folders to your project root:
# From the Maestro-V2 directory, copy these to your project:
cp -r .roo /path/to/your/project/
cp -r personas /path/to/your/project/
cp .roomodes /path/to/your/project/
That's it! The system is ready to use with Roo Code.
- Clone directly into your project root:
cd /path/to/your/project
git clone https://github.com/SannidhyaSah/Maestro-V2.git .maestro-v2
cp -r .maestro-v2/.roo .
cp -r .maestro-v2/personas .
cp .maestro-v2/.roomodes .
If you want to modify the modes or create a fresh configuration:
- Clone this repository:
git clone https://github.com/SannidhyaSah/Maestro-V2.git
cd Maestro-V2
-
Make any changes to the
.roo/
system prompts -
Generate the new Roo configuration:
# For project-specific configuration
node generate-modes.js
# For global configuration
node generate-modes.js --global
- Copy the updated files to your project (see Method 1)
.roo/
- System prompts for all 5 modespersonas/
- Specialized knowledge modules.roomodes
- Roo Code configuration file
Note: The generate-modes.js
script is only needed if you modify the system. For normal use, just copy the folders!
When you begin a task with Maestro, it will ask three questions:
- Is this a new or existing project?
- Complexity rating (1-5)?
- 1-2: Simple tasks routed directly to Prodigy
- 3-5: Complex tasks with full planning workflow
- Agentic mode? (for complexity 3+)
- YOLO: Autonomous mode with minimal questions
- GUIDED: Collaborative mode with user input
User: "Build a React dashboard with authentication"
Maestro: [Analyzes complexity: 4/5]
[Routes to Planner for architecture]
Planner: [Creates technical roadmap]
[Designs component structure]
[Returns to Maestro]
Maestro: [Routes to Coder with React persona]
Coder: [Loads React + Auth personas]
[Implements dashboard]
[Returns to Maestro]
Maestro: [Routes to Code Analyst for review]
Code Analyst: [Loads code-reviewer persona]
[Reviews implementation]
[Provides feedback]
/maestro/
├── /.roo/ # System prompts (Roo reads these)
│ ├── system-prompt-maestro.md
│ ├── system-prompt-code-analyst.md
│ ├── system-prompt-planner.md
│ ├── system-prompt-prodigy.md
│ └── system-prompt-coder.md
├── /personas/ # Specialized knowledge modules
│ ├── /code-analyst/ # Analysis personas
│ └── /coder/ # Development personas
│ ├── /frontend/ # Frontend frameworks & tools
│ ├── /backend/ # Backend languages & frameworks
│ ├── /database/ # Database technologies
│ ├── /testing/ # Testing frameworks
│ ├── /debugging/ # Debugging strategies & tools
│ └── /devops/ # CI/CD & DevOps tools
├── /maestrodocs/ # Runtime workflow management
├── generate-modes.js # Configuration generator
├── LICENSE # Apache 2.0 License
├── README.md # This file
└── CLAUDE.md # AI assistant guidelines
- Mode receives task → Analyzes requirements
- Identifies needed technology → Maps to persona file
- Loads specific persona → Reads only that file
- Applies knowledge → Uses loaded patterns and best practices
- Maintains clean context → No unrelated information loaded
Task: "Create a React component with Material-UI"
1. Maestro → Routes to Coder mode
2. Coder → Identifies: React + Material-UI needed
3. Coder → Loads: /personas/coder/frontend/frameworks/react.md
4. Coder → Loads: /personas/coder/frontend/ui-libraries/material-ui.md
5. Coder → Implements using loaded knowledge
We welcome contributions! Please see our Contributing Guidelines for details.
- System Architecture - Understand how Maestro works
- Adding Personas - Most common contribution
- Modifying Modes - Change mode behavior
- Create a new file in the appropriate category
- Follow the persona template
- Update the mode's persona tree structure
- Submit a pull request
Example: Adding a new frontend framework to Coder mode
- Simplicity: 5 modes is cleaner than 20+
- Modularity: Personas can be added/updated independently
- Efficiency: Load only what's needed
- Clarity: Clear separation of concerns
- Scalability: Easy to add new technologies
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
- Built for Roo Code
- Based on the Maestro V1 workflow system
- Thanks to all contributors and the Roo community
- Issues: GitHub Issues
- Discussions: GitHub Discussions
- Documentation: See CLAUDE.md for AI assistant guidelines
Maestro - Orchestrating AI workflows with precision and clarity