multiagent-task-manager

TaskManager - Multi-Agent Task Management System

A comprehensive JavaScript package for managing tasks across multiple AI agents and human team members. Features intelligent task recommendations, dependency tracking, workload balancing, and seamless collaboration between different types of agents.

🚀 Features

• Multi-Agent Support: Manage both AI agents and human team members
• Intelligent Recommendations: Smart task suggestions based on capabilities and workload
• Dependency Management: Handle complex task dependencies and blocking relationships
• Workload Balancing: Track and optimize task distribution across agents
• Priority & Risk Scoring: Advanced scoring algorithms for task prioritization
• CLI Interface: Command-line tools for quick task management
• MCP Integration: Model Context Protocol support for AI assistants like Claude
• Progress Tracking: Real-time project progress and completion metrics
• Export/Import: Data portability and backup capabilities

📦 Installation

Global Installation

npm install -g multiagent-task-manager

Using npx (No Installation Required)

npx multiagent-task-manager init
npx multiagent-task-manager status
npx task-manager create --title "My task"

Local Project Installation

npm install multiagent-task-manager

🎯 Quick Start

Programmatic Usage

const TaskManager = require('./task-manager.js');

// Initialize
const tm = new TaskManager({ dataDir: './my-project-data' });

// Add agents
tm.addAgent({
  id: 'ai-dev-1',
  name: 'AI Developer',
  type: 'ai',
  capabilities: ['coding', 'testing', 'documentation']
});

tm.addAgent({
  id: 'john.doe',
  name: 'John Doe',
  type: 'human',
  capabilities: ['all']
});

// Create tasks
const task = tm.createTask({
  title: 'Implement user authentication',
  category: 'coding',
  priority: 'high',
  assignees: ['ai-dev-1'],
  completion_criteria: [
    'JWT token generation',
    'Login endpoint',
    'Authentication middleware'
  ]
});

// Get recommendations
const recommendations = tm.getRecommendationsForAgent('ai-dev-1');
console.log('Recommended tasks:', recommendations);

CLI Usage

With npx (recommended - no installation needed)

# Initialize a new project
npx task-manager init

# List all agents  
npx task-manager agents

# Create a new task
npx task-manager create --title "Fix bug" --priority high --assign ai-dev-1

# Get recommendations for an agent
npx task-manager recommend --agent ai-dev-1

# List tasks with filters
npx task-manager list --agent ai-dev-1 --status todo

# Update task status
npx task-manager update TASK-001 --status completed

# Show project status
npx task-manager status

With global installation

# After: npm install -g multiagent-task-manager
task-manager init
task-manager agents
task-manager status

Local development (when working on the package source)

# When developing the package itself
node task-manager.js init
node task-manager.js agents
node task-manager.js status

🔌 MCP Integration (Model Context Protocol)

Connect this task manager to AI assistants like Claude Desktop using MCP.

Quick Setup

# Test access (if installed globally)
multiagent-task-manager-mcp --help

# Add to Claude Desktop config:
{
  "mcpServers": {
    "multiagent-task-manager": {
      "command": "npx",
      "args": ["-y", "--package=multiagent-task-manager", "multiagent-task-manager-mcp"],
      "env": {
        "TASK_MANAGER_DATA_DIR": "~/Documents/TaskManager",
        "TASK_MANAGER_AGENT_ID": "claude-assistant"
      }
    }
  }
}

Available MCP Tools

• init_task_manager - Initialize projects
• create_task - Create and assign tasks
• list_tasks - View and filter tasks
• add_agent - Add team members
• get_recommendations - Get AI suggestions
• get_project_status - View progress
• agent_check_in - Agent workflows
• And 6 more tools...

Example Usage

Ask Claude: "Initialize a new web project with 3 developers and create 5 tasks for a login system"

📚 Full MCP Documentation: See MCP-SETUP.md and QUICK-MCP-SETUP.md

🤖 Agent Management

Agent Types

• AI Agents: Automated agents with specific capabilities
• Human Agents: Human team members with full or specialized capabilities

Agent Structure

{
  id: 'agent-1',
  name: 'Code Assistant',
  type: 'ai', // 'ai' or 'human'
  capabilities: ['coding', 'testing', 'documentation'],
  status: 'active',
  workload: {
    active_tasks: 2,
    completed_tasks: 5,
    total_score: 47
  }
}

Capabilities

• coding: Software development tasks
• testing: Quality assurance and testing
• documentation: Writing and maintaining docs
• design: UI/UX and system design
• analysis: Data analysis and research
• all: Can handle any type of task (typically for humans)

📋 Task Management

Task Structure

{
  id: 'TASK-001',
  title: 'Implement user authentication',
  category: 'coding',
  phase: 'phase-1',
  status: 'todo',
  priority: 'high',
  assignees: [
    {
      id: 'agent-1',
      name: 'Code Assistant',
      type: 'ai',
      role: 'primary',
      assigned_date: '2024-01-15T10:00:00.000Z'
    }
  ],
  dependencies: ['TASK-000'],
  blocks: ['TASK-002'],
  completion_criteria: [
    'JWT token generation',
    'Login endpoint',
    'Authentication middleware'
  ],
  recommendation_score: 23,
  risk_level: 'medium'
}

Task Statuses

• todo: Ready to start
• in-progress: Currently being worked on
• blocked: Waiting for dependencies
• review: Ready for review
• completed: Finished
• cancelled: Cancelled

Priority Levels

• critical: Must be done immediately (🔴)
• high: Important and urgent (🟠)
• medium: Normal priority (🟡)
• low: Can be deferred (🟢)

🎯 Intelligent Recommendations

The system uses advanced scoring algorithms to recommend the best tasks for each agent:

Scoring Factors

• Priority Score: Based on task priority level
• Dependency Score: Higher for tasks that block others
• Risk Score: Prioritizes high-risk tasks
• Phase Score: Active phase tasks get higher scores
• Capability Match: Tasks matching agent capabilities

Getting Recommendations

// Get top 3 recommendations for an agent
const recommendations = tm.getRecommendationsForAgent('agent-1', 3);

recommendations.forEach(task => {
  console.log(`${task.id}: ${task.title}`);
  console.log(`Score: ${task.recommendation_score}`);
  console.log(`Reason: ${task.recommendation_reason}`);
});

🔧 API Reference

TaskManager Class

Constructor

new TaskManager(options)

Options:

• dataDir: Directory for storing data files (default: './tasks-data' with enhanced subdirectory structure)
• maxRecommendations: Maximum recommendations to return (default: 3)
• autoSave: Auto-save changes (default: true)

Agent Methods

// Add a new agent
addAgent(agentInfo)

// Get agent by ID
getAgent(agentId)

// List all agents
listAgents()

// Update agent
updateAgent(agentId, updates)

// Remove agent
removeAgent(agentId)

Task Methods

// Create a new task
createTask(taskData)

// Update task
updateTask(taskId, updates)

// Delete task
deleteTask(taskId)

// Get task by ID
getTask(taskId)

// List tasks with filters
listTasks(filters)

Assignment Methods

// Assign agent to task
assignAgentToTask(taskId, agentInfo)

// Unassign agent from task
unassignAgentFromTask(taskId, agentId)

// Transfer task between agents
transferTask(taskId, fromAgentId, toAgentInfo)

Recommendation Methods

// Get recommendations for agent
getRecommendationsForAgent(agentId, limit)

// Get eligible tasks for agent
getEligibleTasksForAgent(agentId)

Reporting Methods

// Get project status
getProjectStatus()

// Get agent workload
getAgentWorkload(agentId)

🖥️ CLI Commands

Project Management

# Initialization (smart, won't override existing files)
npx task-manager init                        # Initialize with enhanced directory structure
npx task-manager init --current              # Initialize in current directory
npx task-manager init --dir PATH             # Initialize in specific directory

Status and reporting

npx task-manager status # Show project status npx task-manager export # Export project data

### Agent Management
```bash
# Agent management
npx task-manager agents                      # List all agents
npx task-manager agents add                  # Add new agent (interactive)
npx task-manager workload <AGENT_ID>         # Show agent workload

Task Management

# Create tasks
npx task-manager create --interactive               # Interactive task creation
npx task-manager create --title "Task" --priority high --assign agent-1

# List tasks
npx task-manager list                               # List all tasks
npx task-manager list --agent agent-1               # Tasks for specific agent
npx task-manager list --status todo                 # Filter by status
npx task-manager list --priority high               # Filter by priority

# Update tasks
npx task-manager update TASK-001 --status completed
npx task-manager update TASK-001 --priority critical
npx task-manager assign TASK-001 agent-2            # Assign agent to task

Recommendations

npx task-manager recommend --agent agent-1          # Get recommendations

📊 Project Structure

tasks/
├── task-manager.js          # Main package file
├── package.json            # Package configuration
├── README.md               # This documentation
├── examples.js             # Usage examples
├── test-demo.js            # Demo script
├── .env.example            # Environment template
├── src/                    # Enhanced directory utilities
│   ├── constants/          # Path constants and configuration
│   └── utils/              # Directory resolution and safety
└── tasks-data/             # Default data directory (or current dir if --current)
    ├── task-tracker.json   # Task and project data
    ├── agents.json         # Agent registry
    ├── README.md           # Usage guide
    ├── agents/             # Agent-specific files and logs
    ├── reports/            # Generated reports and analytics
    ├── templates/          # Task and agent templates
    └── backups/            # Automated backups

🔄 Workflow Examples

Basic Workflow

• Initialize: Set up the project structure (smart init won't override existing files)
• Configure: Set your agent ID in .env file
• Add Agents: Register AI agents and human team members
• Create Tasks: Define work items with dependencies
• Get Recommendations: Let the system suggest optimal task assignments
• Track Progress: Monitor completion and adjust as needed

Advanced Workflow

// 1. Setup project with multiple agent types
const tm = new TaskManager({ dataDir: './my-project' });

// 2. Add specialized agents
tm.addAgent({
  id: 'backend-ai',
  name: 'Backend AI',
  type: 'ai',
  capabilities: ['coding', 'testing']
});

tm.addAgent({
  id: 'frontend-ai',
  name: 'Frontend AI', 
  type: 'ai',
  capabilities: ['coding', 'design']
});

tm.addAgent({
  id: 'tech-lead',
  name: 'Tech Lead',
  type: 'human',
  capabilities: ['all']
});

// 3. Create task hierarchy with dependencies
const designTask = tm.createTask({
  title: 'System Architecture Design',
  priority: 'critical',
  assignees: ['tech-lead']
});

const backendTask = tm.createTask({
  title: 'API Implementation',
  dependencies: [designTask.id],
  assignees: ['backend-ai']
});

const frontendTask = tm.createTask({
  title: 'UI Implementation',
  dependencies: [backendTask.id],
  assignees: ['frontend-ai']
});

// 4. Monitor and adjust
setInterval(() => {
  const status = tm.getProjectStatus();
  console.log(`Progress: ${status.progress.completion_percentage}%`);
  
  // Get recommendations for available agents
  tm.listAgents().forEach(agent => {
    const recommendations = tm.getRecommendationsForAgent(agent.id);
    if (recommendations.length > 0) {
      console.log(`${agent.name} should work on: ${recommendations[0].title}`);
    }
  });
}, 30000); // Check every 30 seconds

🧪 Testing & Examples

Run the included examples to see the system in action:

# Run all examples (for package development)
node examples.js

# Run demo with sample data (for package development)
node test-demo.js

# Test CLI with demo data (for package development)
TASK_MANAGER_DATA_DIR=./demo-data node task-manager.js status

# For end users, use npx instead:
npx task-manager status
TASK_MANAGER_DATA_DIR=./demo-data npx task-manager status

📁 Enhanced Directory Structure

TaskManager now uses an intelligent directory structure that provides better organization and supports advanced features:

Automatic Directory Detection

The system automatically detects your project context:

• Project Root Detection: Searches for markers like package.json, .git, pyproject.toml
• Current Directory Integration: Uses your current working directory when safe
• Smart Fallback: Multiple fallback strategies for maximum reliability

Directory Layout

your-project/
└── tasks-data/                 # Main TaskManager directory
    ├── task-tracker.json       # Core task and project data
    ├── agents.json             # Agent registry and capabilities
    ├── README.md               # Auto-generated usage guide
    ├── agents/                 # Agent-specific files and configurations
    │   ├── logs/              # Agent activity logs
    │   └── configs/           # Individual agent settings
    ├── reports/               # Generated analytics and reports
    │   ├── workload-report.json
    │   ├── project-status.json
    │   └── recommendations.json
    ├── templates/             # Reusable templates
    │   ├── example-task.json
    │   └── agent-template.json
    └── backups/              # Automated data backups
        └── backup_YYYY-MM-DD_HH-mm-ss/

Directory Safety Features

• System Protection: Prevents writing to unsafe directories (/, /usr/, /opt/, etc.)
• Permission Validation: Checks write permissions before creating directories
• Cross-Platform Support: Works consistently on Windows, macOS, and Linux
• Emergency Fallback: Always provides a safe working directory

Legacy Data Migration

The system automatically migrates data from older formats:

• Detects legacy task-tracker.json and agents.json in project root
• Migrates to new structured format seamlessly
• Preserves all existing task and agent data
• Maintains backward compatibility

Environment Variables

• TASK_MANAGER_AGENT_ID: Your agent identifier for personalized commands
• TASK_MANAGER_DATA_DIR: Override automatic directory detection
• TASK_MANAGER_USE_CURRENT_DIR: Set to 'true' to force current directory usage

📈 Extending the System

Custom Scoring Algorithms

class CustomTaskManager extends TaskManager {
  calculateTaskScore(task) {
    let score = super.calculateTaskScore(task);
    
    // Add custom scoring logic
    if (task.tags && task.tags.includes('urgent')) {
      score += 10;
    }
    
    return score;
  }
}

Custom Agent Capabilities

// Add custom capability checking
tm.getRequiredCapabilities = function(task) {
  const capabilities = [];
  
  if (task.title.includes('ML') || task.title.includes('AI')) {
    capabilities.push('machine-learning');
  }
  
  return capabilities;
};

🤝 Contributing

• Fork the repository
• Create a feature branch
• Add tests for new functionality
• Submit a pull request

📄 License

MIT License - see LICENSE file for details

🆘 Support

For questions and support:

• Check the examples in examples.js
• Review the test scenarios in test-demo.js
• Use the CLI help: npx task-manager

🔮 Future Enhancements

• Web dashboard interface
• Integration with external project management tools
• Advanced analytics and reporting
• Team collaboration features
• Custom notification systems
• API server mode
• Plugin architecture
• Machine learning for better recommendations