AI Collaboration Template

A reusable workflow template for collaborating with AI coding assistants (Claude Code and Codex).

Quick Start

1. Copy this template into your new project repository
2. Customize AGENTS.md with your project-specific technical details
3. Start a session - the AI will follow the workflows automatically

What's Included

your-project/
├── .claude/
│   ├── CLAUDE.md              # Claude Code entry point
│   └── skills/                # Reusable workflow skills
│       ├── commit-push/       # Git workflow automation
│       ├── verify/            # Pre-completion checklist
│       ├── capture-learnings/ # Learning capture workflow
│       ├── refine/            # Meta-prompt optimizer
│       ├── architect/         # Systems design agent
│       ├── backend/           # Backend development agent
│       ├── frontend/          # Frontend development agent
│       ├── reviewer/          # Code review agent
│       └── security/          # Security audit agent
├── WORKFLOWS.md               # Operational guidance (HOW to work)
├── AGENTS.md                  # Technical reference (WHAT the project is)
├── tasks/
│   └── lessons.md             # Learning log
└── README.md                  # This file

Philosophy

This template establishes a collaboration model where:

• You are the Product Owner - you make decisions
• AI is your Technical Co-Founder - handles building, keeps you informed
• Learning is built-in - explanations, options, and knowledge capture
• Quality is non-negotiable - real products, not prototypes

Customization

Required Customization

1. AGENTS.md - Fill in [PROJECT-SPECIFIC] sections with your:
   • Project overview and architecture
   • Tech stack
   • Repository structure
   • API endpoints (if applicable)
   • Database schema (if applicable)
   • Environment variables

Optional Customization

• WORKFLOWS.md - Adjust thresholds, add project-specific patterns
• Skills - Add domain-specific skills as needed
• tasks/lessons.md - Will evolve naturally as you work

How It Works

Session Start

AI reads lessons.md → WORKFLOWS.md → AGENTS.md → understands context

During Work

AI follows 5-phase framework → explains as it goes → captures learnings

Session End

AI updates lessons.md → suggests improvements → documents progress

Model Routing

Model	Role	Use For
Opus (latest)	Brains	Planning, architecture, complex decisions
Sonnet 4.5	Hands	Implementation, editing, routine tasks

License

Use freely. Adapt to your needs.