AutoClaude is a cross-platform tool that enables multi-AI agent collaboration through a file-based task system, supporting 24/7 continuous development with automatic quality control.
v3.0.0-beta.1 introduces major improvements to unify all entry points:
- Unified Node.js Actions: Right-click context menus now use Node.js scripts instead of PowerShell
- Cross-Platform Context Menus: Better support for Windows, with preparations for macOS/Linux
- Enhanced CLI: New
autoclaude context-menu install/uninstallcommands - Proper Daemon Management:
autoclaude stopnow works with PID file tracking - Better Error Handling: System notifications for all actions
- Right-click menus now call Node.js scripts in
src/actions/ - PowerShell scripts moved to legacy reference in
scripts/ - Context menu registration improved with better Node.js path detection
If upgrading from v2.x:
- Uninstall old context menus:
autoclaude context-menu uninstall - Reinstall:
autoclaude context-menu install - All functionality remains the same, just more reliable!
AutoClaude creates a collaboration system where multiple AI agents work together:
┌─────────────────────────────────────────────────────────────────┐
│ You: Write requirements in collaboration/inbox.md │
└─────────────────────────────────────────────────────────────────┘
↓
┌─────────────────────────────────────────────────────────────────┐
│ Supervisor: Analyzes requirements → Creates task files │
└─────────────────────────────────────────────────────────────────┘
↓
┌─────────────────────────────────────────────────────────────────┐
│ Executor: Picks up task → Writes code → Runs tests → Submit │
└─────────────────────────────────────────────────────────────────┘
↓
┌─────────────────────────────────────────────────────────────────┐
│ Supervisor: Reviews critically → Approves or requests changes │
└─────────────────────────────────────────────────────────────────┘
↓
┌─────────────────────────────────────────────────────────────────┐
│ Final: High-quality, tested, well-structured code │
└─────────────────────────────────────────────────────────────────┘
- Windows, macOS, Linux: Native support for all platforms
- Node.js powered: No PowerShell or shell script dependencies
- npm installable: Global CLI tool available anywhere
# Install globally
npm install -g autoclaude
# Initialize any project
cd your-project
autoclaude init
# Start the watcher
autoclaude start- Supervisor: Breaks down requirements into actionable tasks
- Executor: Implements tasks with testing and documentation
- Watcher: Monitors files and coordinates agent interactions
- Quality Control: Multiple review cycles ensure high-quality output
# Create a task from command line
autoclaude task "Build a user authentication system"
# Check project status
autoclaude status
# View dashboard
autoclaude dashboard- Self-testing: Executors write tests for every implementation
- Continuous validation: Test suite runs automatically
- Coverage tracking: Monitors test coverage and quality
npm install -g autoclaudenpm install autoclaude
npx autoclaude initgit clone https://github.com/AcePeak/AutoClaude.git
cd AutoClaude
npm install
npm link # Makes `autoclaude` available globallycd your-project
autoclaude initThis creates:
collaboration/directory with task management structurecollaboration/inbox.mdfor writing requirements- Configuration files and templates
Edit collaboration/inbox.md and describe what you want built:
# Requirements
Build a todo application with the following features:
- Add, edit, and delete tasks
- Mark tasks as complete
- Filter by status (all, active, completed)
- Persist data to localStorageautoclaude startThe watcher will:
- Detect new requirements in inbox.md
- Trigger the Supervisor to create task files
- Executors will implement each task
- Supervisor will review and approve completed work
autoclaude init [directory] # Initialize a project
autoclaude start [directory] # Start the watcher
autoclaude stop # Stop the watcher daemon
autoclaude status [directory] # Show project statusautoclaude task "description" # Create a new task
autoclaude dashboard # Open project dashboard
autoclaude projects # List all registered projectsautoclaude watch --once # Run one check cycle
npm test # Run tests
npm run test:coverage # Run tests with coverageAfter initialization, your project will have:
your-project/
├── collaboration/
│ ├── inbox.md # Write requirements here
│ ├── queue/ # Tasks waiting to be executed
│ ├── executing/ # Tasks currently being worked on
│ ├── completed/ # Finished tasks
│ ├── SUPERVISOR_GUIDE.md # Instructions for the Supervisor
│ ├── EXECUTOR_GUIDE.md # Instructions for Executors
│ └── .autoclaude/
│ ├── config.json # Project configuration
│ ├── logs/ # Execution logs
│ └── lock/ # Process lock files
├── src/ # Your source code
└── tests/ # Generated tests
Edit collaboration/.autoclaude/config.json:
{
"name": "My Project",
"max_executors": 2,
"check_interval_seconds": 60,
"claude_path": "claude"
}AutoClaude stores global settings in your OS-appropriate config directory:
- Windows:
%APPDATA%\\AutoClaude\\settings.json - macOS:
~/Library/Application Support/AutoClaude/settings.json - Linux:
~/.config/autoclaude/settings.json
AutoClaude uses a file-based system for agent communication:
- Requirements →
collaboration/inbox.md - Task Queue →
collaboration/queue/*.md - In Progress →
collaboration/executing/*.md - Completed →
collaboration/completed/*.md
- Monitors file changes
- Triggers Supervisor when inbox has content
- Triggers Executors when queue has tasks
- Manages orphaned tasks and cleanup
- Analyzes requirements in inbox.md
- Breaks down complex requirements into specific tasks
- Reviews completed work for quality
- Decides approval/rejection with feedback
- Claims tasks from the queue
- Implements features with proper testing
- Submits completed work for review
- Handles task recovery and resumption
# List all registered projects
autoclaude projects
# Work with specific project
autoclaude status /path/to/project
autoclaude start /path/to/projectYou can customize task templates by modifying the Supervisor guide:
vim collaboration/SUPERVISOR_GUIDE.md
vim collaboration/EXECUTOR_GUIDE.md# Start as background daemon
autoclaude start --daemon
# Check if running
ps aux | grep autoclaudeAutoClaude includes a comprehensive test suite:
# Run all tests
npm test
# Run with coverage
npm run test:coverage
# Run specific tests
npm test -- tests/core/watcher.test.js
# Watch mode during development
npm run test:watch- Fork the repository
- Create a feature branch
- Write tests for your changes
- Ensure all tests pass
- Submit a pull request
git clone https://github.com/AcePeak/AutoClaude.git
cd AutoClaude
npm install
npm test- src/core/: Core agent implementations (Watcher, Supervisor, Executor)
- src/cli/: Command-line interface
- src/utils/: Utility functions and helpers
- tests/: Comprehensive test suite
Ensure the Claude CLI is installed and available in your PATH:
which claude # Should return a path
claude --version # Should show versionOn Unix systems, ensure the CLI script is executable:
chmod +x node_modules/.bin/autoclaudeCheck for orphaned lock files:
autoclaude status # Shows current task statesCheck execution logs for detailed information:
ls collaboration/.autoclaude/logs/
tail -f collaboration/.autoclaude/logs/supervisor_*.logAutoClaude v3.x is fully cross-platform and doesn't require PowerShell. Existing Windows projects should continue working, but for best compatibility:
- Install via npm:
npm install -g autoclaude - Use
autoclaude startinstead of PowerShell scripts - Update any custom scripts to use Node.js instead of PowerShell
MIT License - see LICENSE for details.
- 🐛 Issues
- 💡 Discussions
- 💖 Sponsor
AutoClaude: Where AI agents collaborate to build software, so you don't have to.