Multi-Agent Collaborative Analysis

Spawn 3 AI agent team members (NOT subagents) with distinct analytical perspectives. They communicate with each other via SendMessage, debate trade-offs, then present a unified comparison for the user to choose from.

Prerequisites

Agent teams are experimental. Ensure the feature is enabled:

// settings.json
{
  "env": {
    "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
  }
}

Critical: Team Members vs Subagents