🌀 Tazz CLI

AI-powered task orchestrator with git worktrees, tmux sessions, and MCP integration

Tazz CLI is a sophisticated development orchestrator that combines the power of AI agents (Claude) with robust session management, git worktrees, and intelligent codebase analysis to create isolated, resumable development environments.

✨ Features

🧠 Intelligent Initialization

• Automatic codebase analysis - Understands your project structure, technologies, and patterns
• MCP Integration - Seamlessly connects with your existing Claude Code MCP servers
• Smart rules generation - Creates project-specific coding standards and quality gates
• Testing strategy setup - Configures appropriate test templates and coverage targets

🌪️ Session Management

• Git worktree isolation - Each task gets its own isolated git worktree
• Tmux integration - Multi-pane terminal sessions for organized development
• Resumable sessions - Sessions survive reboots and can be resumed anytime
• Context-aware environments - Sessions remember their state and context

🔗 MCP & Jira Integration

• Jira ticket integration - Automatically fetches ticket context and creates tasks
• GitHub PR management - Tracks related pull requests and feedback
• SonarCloud quality gates - Integrates code quality metrics and analysis
• Multi-agent orchestration - Coordinate multiple AI agents across sessions

🎯 Development Workflow

• Tornado animations - Delightful CLI experience with custom animations
• Quality gates - Automated testing, linting, and coverage checks
• Hook system - Claude Code hooks for automated workflows
• Task management - Markdown-based task tracking with agent integration

🚀 Quick Start

Installation

npm install -g tazz-cli

Initialize Tazz in your project

cd your-project
tazz make

This will:

• Analyze your codebase and detect technologies
• Set up MCP integration with your Claude Code configuration
• Generate intelligent rules and coding standards
• Create specialized AI agents (validation-engineer, software-architect, e2e-engineer, etc.)
• Copy complete Claude Code configuration with subagents
• Configure testing strategies and templates
• Create hook scripts for automated workflows

Start a new development session

# For Jira tickets (auto-fetches context)
tazz start JIRA-123

# For general features
tazz start feature-authentication

# With custom tasks
tazz start bugfix-login --tasks "fix auth bug, add tests, update docs"

Manage sessions

# List all sessions
tazz list

# Attach to existing session
tazz attach JIRA-123

# Stop session (keeps worktree)
tazz stop JIRA-123

# Delete session and worktree
tazz delete JIRA-123

Detached console mode

# Start interactive tornado console
tazz -d

🛠️ Configuration

Tazz automatically configures itself based on your project. Configuration is stored centrally in ~/.tazz/projects/<project-hash>/config.json while project-specific settings are in .claude/settings.json:

{
  "maxConcurrentSessions": 10,
  "defaultBranch": "main",
  "tmuxPrefix": "tazz_",
  "qualityGates": {
    "enabled": true,
    "coverage": 80
  },
  "agents": {
    "claude": {
      "enabled": true,
      "model": "claude-3-sonnet-20240229"
    }
  }
}

🔧 MCP Integration

Tazz automatically detects and integrates with your existing Claude Code MCP servers:

• Git MCP - Repository analysis and operations
• Atlassian MCP - Jira ticket integration
• GitHub MCP - Pull request management
• SonarCloud MCP - Code quality analysis
• Playwright MCP - E2E test generation

📁 Project Structure

After initialization, Tazz creates:

your-project/
├── .tazz/
│   └── tazz-todo.md        # User collaboration file (only file in .tazz)
├── .claude/                # Complete Claude Code configuration
│   ├── agents/            # Specialized AI agent configurations
│   │   ├── validation-engineer.md     # Testing & QA specialist
│   │   ├── software-architect.md      # Architecture & design expert
│   │   ├── e2e-engineer.md           # End-to-end testing specialist
│   │   ├── devops-engineer.md        # CI/CD & infrastructure expert
│   │   ├── frontend-engineer.md      # Frontend development specialist
│   │   └── backend-engineer.md       # Backend development specialist
│   ├── code-quality-rules.md         # Code quality standards
│   ├── engineering-practices.md      # Development practices
│   ├── git-workflow-rules.md         # Git workflow guidelines
│   ├── quality-gates.md              # Quality gate definitions
│   ├── settings.json                 # Claude Code settings (includes MCP)
│   ├── settings.local.json           # Local overrides
│   ├── hooks/                        # Claude Code hook scripts
│   └── rules/                        # Generated coding rules
├── ~/.tazz/projects/<project-hash>/  # Centralized configuration
│   ├── config.json                   # Tazz configuration
│   ├── sessions.json                 # Active sessions
│   ├── analysis.json                 # Codebase analysis results
│   ├── templates/                    # Test templates
│   └── logs/                         # Session logs
└── .gitignore                        # Updated to ignore Tazz files

🤖 Specialized AI Agents

Tazz includes pre-configured specialized AI agents that are automatically set up in your .claude/agents/ directory:

🧪 Validation Engineer

• Focus: Testing strategies, test automation, QA processes
• Expertise: TDD, BDD, unit/integration/E2E testing, test coverage
• Tools: Jest, Vitest, Playwright, Cypress, testing frameworks

🏗️ Software Architect

• Focus: System design, architectural patterns, technical decisions
• Expertise: SOLID principles, DDD, Clean Architecture, CQRS, design patterns
• Tools: Architecture analysis, code structure optimization

🌐 E2E Engineer

• Focus: End-to-end testing, user workflow validation
• Expertise: Playwright, cross-browser testing, accessibility testing
• Tools: Browser automation, visual regression testing

🚀 DevOps Engineer

• Focus: Infrastructure automation, CI/CD pipelines, deployment
• Expertise: Docker, Kubernetes, Terraform, GitHub Actions, monitoring
• Tools: Infrastructure as Code, deployment strategies

💻 Frontend Engineer

• Focus: Modern web development, UI/UX implementation
• Expertise: React/Vue/Angular, state management, performance optimization
• Tools: Component libraries, build tools, responsive design

⚙️ Backend Engineer

• Focus: Server-side development, APIs, databases
• Expertise: Node.js/Express, database design, authentication, microservices
• Tools: API development, database optimization, security

Agent Commands

# Run tasks in parallel across multiple sessions
tazz agent run --tasks "fix bug, write tests, update docs" --parallel

# Target specific sessions
tazz agent run --tasks "review code" --sessions "JIRA-123,JIRA-456"

🎯 Advanced Usage

Custom Branch Naming

tazz start JIRA-123 --branch feature/JIRA-123-custom-auth

Skip Integrations During Init

tazz make --skip-mcp --skip-analysis --skip-hooks

Force Reinitialize

tazz make --force

🧪 Testing

Tazz includes comprehensive testing with:

• Unit tests with Vitest
• Integration tests for MCP services
• E2E tests for CLI workflows
• 90%+ code coverage target

npm test
npm run test:coverage
npm run test:e2e

🚢 Development

Build from source

git clone https://github.com/tazz-dev/tazz-cli.git
cd tazz-cli
npm install
npm run build
npm link

Project Architecture

src/
├── cli/                    # CLI commands and UI
├── core/
│   ├── services/          # Business logic
│   ├── storage/           # Data persistence
│   └── types/             # TypeScript definitions
├── agents/                # AI agent integration
└── utils/                 # Utilities and helpers

📋 Requirements

• Node.js 18+
• Git (for worktree support)
• tmux (for session management)
• Claude Code (for MCP integration)

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

📄 License

MIT © Tazz Development Team

🙏 Acknowledgments

• Anthropic for Claude and the MCP ecosystem
• Claude Code for hooks and development environment integration
• The open-source community for the excellent tools and libraries

Website • Documentation • Issues • Discord