Skip to content

CLAUDE.md

Latest commit

 

History

History
134 lines (105 loc) · 4.97 KB

CLAUDE.md

File metadata and controls

134 lines (105 loc) · 4.97 KB
Raw

CLAUDE.md

Welcome to the 3 AI Amigos workspace - a complete toolkit for orchestrating AI-powered Product Managers, UX Designers, and Claude Code to build production systems.

🎯 What You're Looking At

This repository contains everything needed to implement the 3 AI Amigos methodology:

  • Agent configurations to create your AI team
  • Technical patterns that work for any domain
  • Complete example of a multi-agent health system
  • Setup guides to get you started

🚀 Quick Navigation

Want to implement 3 AI Amigos? Start here:

  1. README.md - Your complete guide (start here!)
  2. agents/ - Copy these configs to set up your AI agents
  3. docs/po-document-checklist.md - Documents you'll need

Want to understand the patterns?

  • technical-guides/implementation-guide.md - How to build multi-agent systems
  • technical-guides/multi-agent-patterns.md - Orchestrator-worker architecture
  • docs/images/ai-amigos-diagram.png - Visual overview

Want to see a complete example?

  • use-cases/multi-agent-health-insight-system/ - Full health system implementation
    • Includes brand guidelines based on the mockups
    • Complete domain documentation
    • Visual examples in the PDF

📁 Repository Structure

3-AMIGO-AGENTS/
│
├── 🤖 agents/                    # COPY THESE to configure your agents
│   ├── pm-agent/                 
│   │   ├── config/              # PM Agent instructions
│   │   └── example-prompts/     # What to say to PM
│   ├── ux-agent/                
│   │   ├── config/              # UX Agent instructions  
│   │   └── example-prompts/     # What to say to UX
│   └── code-agent/              
│       └── config/              # CLAUDE.md template for your workspace
│
├── 📚 technical-guides/          # REUSABLE patterns (domain-agnostic)
│   ├── implementation-guide.md   # Multi-agent implementation
│   ├── multi-agent-patterns.md   # Based on Anthropic's research
│   ├── multi-agent-implementation-architecture.md # Backend structure patterns
│   └── streaming-patterns.md     # Real-time updates
│
├── 🏥 use-cases/                 # EXAMPLE: Complete health system
│   └── multi-agent-health-insight-system/
│       ├── [domain documents]    # Templates for your own domains
│       └── health-user-stories.pdf # UI/UX examples
│
├── 📖 docs/                      # HOW-TO guides
│   ├── demo-setup-guide.md      # Run the demo
│   ├── po-document-checklist.md # What docs you need
│   └── images/                  # Diagrams and visuals
│
└── README.md                     # START HERE - Complete guide

💡 The Big Picture

The 3 AI Amigos transforms development by orchestrating:

  1. Product Manager Agent → Creates requirements & architecture
  2. UX Designer Agent → Designs interfaces & experiences
  3. Claude Code → Implements the complete system

You provide domain expertise, they build the system.

🎯 Your Next Steps

Option 1: Build Something

  1. Read README.md thoroughly
  2. Set up your 3 AI agents using files in agents/*/config/
  3. Create domain documents (use health example as template)
  4. Run through the 5-phase process
  5. Get a working system!

Option 2: Explore First

  1. Look at docs/images/ai-amigos-diagram.png
  2. Browse the health example in use-cases/
  3. Read technical guides to understand patterns
  4. Then follow Option 1 when ready

🔑 Key Files You'll Need

To Configure Agents:

  • agents/pm-agent/config/pm-agent-instructions.md
  • agents/ux-agent/config/ux-designer-agent-instructions.md
  • agents/code-agent/config/minimal-claude-md-template.md

To Understand What to Submit:

  • agents/pm-agent/example/prompts/po-prompt-for-pm-agent.md - Prompt + attached docs
  • agents/ux-agent/example/prompts/po-prompt-for-ux-agent.md - Prompt + attached docs

To See What Documents You Need:

  • docs/po-document-checklist.md
  • use-cases/multi-agent-health-insight-system/ (examples)

📝 Important Notes

  1. Domain-Agnostic vs Domain-Specific

    • agents/ and technical-guides/ work for ANY domain
    • use-cases/ shows specific examples
    • Keep them separate!

  2. The Process is Sequential

    • PM Agent first → UX Agent second → Claude Code last
    • Each builds on the previous outputs

  3. You're the Product Owner

    • You provide domain knowledge
    • You guide the agents
    • You validate outputs

🎉 Success Looks Like

  • ✅ 3 AI agents configured in Claude Desktop
  • ✅ Your domain requirements ready
  • ✅ PM creates comprehensive specs
  • ✅ UX creates beautiful designs
  • ✅ Claude Code builds working system
  • ✅ System runs successfully!

Ready? Open README.md and let's build something amazing with your new AI development team!

Remember: This isn't just about tools - it's about revolutionizing how we build software. 🚀