Baryon Agents Parallel Workspace | Spec-driven AI Agents Parallel Isolated Workspace

🚀 Integrated Environment for Agents Parallel Programming - Provides simultaneous workspace for Claude, Cursor, Gemini, Kino agents with Git worktree isolation, TMUX session management, and flexible Spec-driven workflow conversion.

🎪 Core Concepts

🤖 Agents Parallel Programming: Multiple AI agents working simultaneously and independently 🔄 Spec-Driven Workflow Conversion: Flexible conversion from master specification to agent-specific formats (Claude MD, Cursor MDC, Gemini MD, Kino AWS) 🌿 Git Worktree Isolation: Independent branches and workspaces for each agent 🖥️ TMUX Session Management: Stable terminal environment and session persistence 📊 Real-time Monitoring: Real-time tracking of agent activities and status

🚀 Overview

🚀 Integrated Environment for Agents Parallel Programming - Complete development ecosystem for simultaneous AI agent collaboration This extension implements an industry-first approach to multi-agent AI development by:

Converting a single master specification into multiple AI agent formats automatically
Providing isolated Git worktree environments for each agent (inspired by Claude Squad)
Enabling real-time synchronization and collaboration between AI agents
Offering enterprise-grade observability and monitoring capabilities

✨ Key Features

🎯 Industry-First Capabilities

Single Spec → Multi-Agent Format Conversion: Write once, deploy to Claude, Cursor, Gemini, and Kino
Git Worktree Isolation: Each agent operates in its own isolated environment
Unified Dev Server: Single port (5174) with path-based routing for all worktrees
Claude Code Sub Agents: Specialized AI assistants with MCP pair-role patterns
Parallel Execution Engine: Advanced parallel AI coding with intelligent scheduling
Real-time Synchronization: Automatic format conversion and state synchronization
VSCode Integration: Native UI dashboard vs terminal-based competitors
Claude Hook Manager: Automatic installation and management of Claude CLI hooks

📊 Competitive Advantages

2x faster workflows compared to Claude Squad
5x improved collaboration compared to Kiro.dev
Exclusive real-time sync not available in any competitor
Enterprise-ready with comprehensive observability

🛠️ Installation

System Requirements

Optimized for macOS 🍎 (Full feature support)
Windows and Linux: Basic functionality (TMUX features may be limited)
TMUX required for advanced terminal management
- macOS: brew install tmux
- Linux: Use your package manager (apt, yum, etc.)
- Windows: Use WSL2 with TMUX

Installation Steps

Install from VSCode Marketplace (coming soon)

Or install manually:

git clone https://github.com/your-repo/multi-agent-spec-vscode
cd multi-agent-spec-vscode
npm install
npm run build

🚀 Quick Start

🎯 Enhanced with Claude Agents Power

Install Claude Agents Power (Recommended)
```
npm install -g claude-agents-power
```
Initialize 100+ Specialized Agents
```
claude-agents-power --install
```

Ask Claude to Set Up Your Team

"claude-agents-power를 이용해서 이 프로젝트를 위한 추천 및 agents/*.md 파일을 다운로드 해줘"

📋 Basic Setup

Initialize Project

Cmd/Ctrl + Shift + P → AI Multi-Agent: Initialize Project

Create Master Specification
- Edit .ai-specs/master-spec.md with your requirements
- Use EARS format for requirements
- Define design sections and tasks
Configure Agents
- Edit .ai-specs/config.json to enable desired agents
- Set agent-specific configurations

Launch Agents

Cmd/Ctrl + Shift + P → AI Multi-Agent: Create Instance

⚙️ Configuration

Extension Settings

Configure the extension through VSCode settings:

{
  // Auto-initialize Claude hooks on extension activation (default: true)
  "aiMultiAgent.autoInitializeClaudeHooks": true,
  
  // Enable automatic synchronization
  "aiMultiAgent.enableAutoSync": true,
  
  // Sync debounce time in milliseconds
  "aiMultiAgent.syncDebounceMs": 1000,
  
  // Default agent types to create
  "aiMultiAgent.defaultAgentTypes": ["claude", "cursor"],
  
  // Maximum number of git worktrees
  "aiMultiAgent.maxWorktrees": 10
}

Claude Hook Management

The extension now includes integrated Claude hook management:

Automatic Installation: Hooks are installed automatically when the extension activates
Global Management: Hooks are installed in ~/.claude/ for system-wide use
Safe Operation: Extension-managed hooks are marked and can be safely removed
Commands Available:
- Initialize Claude Hooks: Manually install hooks
- Check Claude Hooks Status: View current hook status
- Toggle Claude Hooks: Enable/disable hooks
- Uninstall Claude Hooks: Remove extension-managed hooks

🔔 Agent Activity Monitoring (NEW!)

The extension now includes a powerful Agent Activity monitoring system that tracks and displays real-time notifications from Claude Desktop and other AI agents.

Features

Real-time Activity Feed: See what your AI agents are doing in real-time
Smart Categorization: Activities are automatically categorized by type
Sound Notifications: Optional sound alerts for important activities
Activity Grouping: Group by time, agent type, or project
Detailed Metadata: Click any activity to see full details
Claude Hook Integration: Automatic management of Claude CLI hooks for enhanced observability

Setup Claude Desktop Integration

Automatic Setup (Recommended)

The extension automatically manages Claude hooks for you. No manual installation needed!

Manual Setup (Optional)

If you prefer manual control:

Install hooks: Cmd/Ctrl + Shift + P → AI Multi-Agent: Initialize Claude Hooks
Check status: Cmd/Ctrl + Shift + P → AI Multi-Agent: Check Claude Hooks Status
Toggle hooks: Cmd/Ctrl + Shift + P → AI Multi-Agent: Toggle Claude Hooks
View activities in VSCode:
- Open the sidebar
- Click the "Baryon Agents Parallel Workspace" icon
- Expand "Agent Activities"

Test the integration:

./scripts/claude-hooks/test-vscode-integration.sh

What Gets Tracked

📖 File reads and edits
🖥️ Command executions
🔍 Search operations
💬 User prompts (categorized by type)
📊 Conversation summaries
🔧 Tool usage with context

Configuration

{
  "aiMultiAgent.hookListenerPort": 8765,
  "aiMultiAgent.enableNotificationSound": true
}

For detailed setup instructions, see Claude Hook Integration Guide.

📁 Project Structure

your-project/
├── .ai-specs/
│   ├── master-spec.md      # Master specification
│   ├── config.json         # Extension configuration
│   ├── logs/              # Sync and error logs
│   └── worktrees/         # Agent worktree metadata
├── .claude/               # Claude agent format
├── .cursor/               # Cursor agent format
├── .gemini/               # Gemini agent format
├── .kiro/                 # Kiro agent format
└── trees/                 # Git worktrees for parallel development
    ├── claude-feature-1/
    ├── cursor-feature-2/
    └── gemini-feature-3/

🔧 Configuration

Basic Configuration

{
  "agents": {
    "claude": {
      "enabled": true,
      "format": ".claude/rules.md"
    },
    "cursor": {
      "enabled": true,
      "format": ".cursor/rules.mdc"
    }
  },
  "sync": {
    "debounceMs": 1000,
    "retryAttempts": 3
  }
}

📋 Commands

Command	Description
`aiMultiAgent.simpleInitParallel`	Initialize parallel worktrees for multiple agents
`aiMultiAgent.initialize`	Initialize project with template
`aiMultiAgent.showDashboard`	Open agent dashboard
`aiMultiAgent.createPullRequest`	Create PR from current worktree
`aiMultiAgent.applyToMain`	Apply worktree changes to main branch
`aiMultiAgent.syncWithMain`	Sync worktree with latest main
`vscode-baryon.manageTelemetry`	Manage telemetry settings

🔍 Agent Observability

The extension provides comprehensive monitoring:

Real-time agent status (Running/Idle/Stopped/Error)
Activity history and event logging
Performance metrics and analytics
Collaboration insights between agents

🏗️ Architecture

The extension follows a Model-View-Controller pattern:

Extension Controller: Main coordinator
Specification Manager: Master spec management
Agent Manager: Lifecycle management
Synchronization Engine: Format conversion
Worktree Manager: Git isolation
UI Provider: VSCode integration

🤖 Claude Code Sub Agents

The extension includes specialized Claude Code Sub Agents with MCP pair-role patterns for enhanced collaboration:

🆕 MCP Pair-Role Agents (NEW!)

architect-reviewer: Design proposal + immediate validation in one workflow
implementer-tester: TDD implementation with real-time testing
explorer-optimizer: Innovation exploration + performance optimization

Available Sub Agents

vscode-extension-architect: VSCode extension architecture and design
typescript-expert: Advanced TypeScript patterns and type safety
git-worktree-specialist: Git worktree management and optimization
vscode-test-runner: Test execution and failure fixing
spec-validator: Master specification validation
performance-optimizer: Performance analysis and optimization
security-auditor: Security vulnerability scanning

Usage Example

# MCP pair-role for comprehensive design
> Use architect-reviewer to design the new parallel execution feature

# Traditional single agent
> Use vscode-extension-architect to design the new feature

# Chain multiple agents
> First use spec-validator to check the spec, then use vscode-test-runner to verify tests

See Claude Sub Agents Guide for detailed information.

🌐 Unified Dev Server

Access all worktrees through a single port with path-based routing:

# Dashboard
http://localhost:5174/

# Individual worktrees
http://localhost:5174/repo/main
http://localhost:5174/repo/feature-auth
http://localhost:5174/repo/feature-ui

Benefits:

No port conflicts
Unlimited worktree scaling
Centralized management
Automatic server lifecycle

🚧 Roadmap

Phase 1 ✅ (Completed)

Master specification models
EARS format validation
Configuration management
Git worktree functionality
Format transformation engine
Real-time synchronization
VSCode UI dashboard
Unified Dev Server
Claude Code Sub Agents

🖥️ Platform Support

Platform	Support Level	Features	Notes
macOS	✅ Full Support	All features including TMUX	Primary development platform
Linux	⚡ Good Support	Most features work	TMUX installation required
Windows	⚠️ Basic Support	Core features only	Use WSL2 for TMUX support

Known Limitations

Windows: Native TMUX not supported, use WSL2
Linux: Some distributions may require additional setup
All platforms: Git 2.20+ required for worktree support

🚨 Troubleshooting

Having issues? Check our comprehensive Troubleshooting Guide for:

Quick fixes for common problems
Step-by-step solutions
Platform-specific help
Advanced debugging tips

Quick Fixes

Git not found? Install from git-scm.com
Permission denied? Run chmod 755 trees/
Worktree exists? Use cleanup command: "A3 Agent Spec: Cleanup Inactive Workspaces"
Extension not loading? Reload window: Cmd+Shift+P → "Developer: Reload Window"

💬 Support

Troubleshooting Guide: TROUBLESHOOTING_GUIDE.md
Report issues at GitHub Issues
Documentation at Project Wiki
Discussion at GitHub Discussions
Email: support@baryon.ai

🌟 Inspired By

This project draws inspiration from several innovative AI development tools:

Claude Squad - Git worktree-based agent isolation concept
Kiro.dev - Multi-agent collaboration workflows
Cursor - AI-powered code editor integration
Gemini Code Assist - Enterprise AI coding assistance

We've combined the best ideas from these tools to create a unified, VSCode-native experience for parallel multi-agent development.

Built with ❤️ for the AI development community

Baryon Agents Parallel Workspace | Spec-driven AI Agents Parallel Isolated Workspace

Baryon AI Labs

Baryon Agents Parallel Workspace | Spec-driven AI Agents Parallel Isolated Workspace

🎪 Core Concepts

🚀 Overview

✨ Key Features

🎯 Industry-First Capabilities

📊 Competitive Advantages

🛠️ Installation

System Requirements

Installation Steps

🚀 Quick Start

🎯 Enhanced with Claude Agents Power

📋 Basic Setup

⚙️ Configuration

Extension Settings

Claude Hook Management

🔔 Agent Activity Monitoring (NEW!)

Features

Setup Claude Desktop Integration

Automatic Setup (Recommended)

Manual Setup (Optional)

What Gets Tracked

Configuration

📁 Project Structure

🔧 Configuration

Basic Configuration

📋 Commands

🔍 Agent Observability

🏗️ Architecture

🤖 Claude Code Sub Agents

🆕 MCP Pair-Role Agents (NEW!)

Available Sub Agents

Usage Example

🌐 Unified Dev Server

🚧 Roadmap

Phase 1 ✅ (Completed)

🖥️ Platform Support

Known Limitations

🚨 Troubleshooting

Quick Fixes

💬 Support

🌟 Inspired By