Team Skill

Spawns parallel workers to tackle tasks concurrently.

When to Use

• •User says "team", "swarm", "parallel", "coordinated team"
• •Multiple independent tasks
• •Need to work on different parts simultaneously
• •Time-critical execution

When NOT to Use

• •Single sequential task
• •Tasks have tight dependencies
• •Simple quick task

Usage

omk team [ralph] <N>:<agent-type> "<task description>"

Examples:

• •omk team 3:executor "find and fix bugs"
• •omk team ralph 2:architect "design API"

How It Works

1. •Parse args: Extract worker count, agent type, task
2. •Create state: Set up team directory structure
3. •Divide work: Split task into subtasks
4. •Spawn workers: Each worker gets a subtask
5. •Coordinate: Workers report progress
6. •Complete: Collect results

Team Structure

.kimiplugin/state/team/{team-name}/
├── config.json       # Team configuration
├── workers/
│   ├── worker-0/
│   │   ├── config.json   # Worker task
│   │   └── result.json   # Worker output
│   └── worker-1/
│       └── ...

Commands

• •omk team 3:executor "task" - Start team
• •omk team status <name> - Check status
• •omk team shutdown <name> - Stop team

Best Practices

• •Divide tasks clearly
• •Minimize dependencies between workers
• •Define clear success criteria
• •Use appropriate agent types