ArchitectureClaude Code

Multi-Agent Task Decomposition

Break a complex task into pieces for parallel AI agents

Prompt

I want to implement [describe complex feature]. Design a multi-agent approach: 1. What independent subtasks can run in parallel? 2. What shared context does each agent need? 3. What are the integration points between agents' outputs? 4. What verification should happen after agents complete? 5. What's the coordination strategy — fan-out/fan-in, pipeline, or swarm? I'm using [Claude Code/Cursor] with [describe available agent capabilities].