AI

Personal tools, configs, and guidelines for AI development.

Contents

• AGENTS.md - Enterprise-grade agent development guide
  • Code quality standards
  • Architecture patterns
  • Testing requirements
  • Security best practices
  • Performance optimization
  • ADR workflow for architectural decisions
• AGENT-WORKSPACE.md - Workspace-specific template
  • Project context and setup
  • Development commands
  • Current work tracking
  • Project-specific patterns

Purpose

Reference materials and configurations for building production AI systems.

Installation

Quick Install (Recommended)

Install via npx (no download required):

npx @sethwebster/ai-cli

Or install globally via npm:

npm install -g @sethwebster/ai-cli

Then reload your shell and initialize your project:

source ~/.zshrc  # or ~/.bashrc for bash
ai init

Alternative: Direct Install

Install with curl (no Node.js required):

curl -fsSL https://raw.githubusercontent.com/sethwebster/AI/main/install.sh | bash

Manual Installation

Add the function to your shell config:

# Download the function
curl -fsSL https://raw.githubusercontent.com/sethwebster/AI/main/ai-function.sh >> ~/.zshrc

# Reload your shell
source ~/.zshrc

# Initialize a project
ai init

Structure

AI/
├── agents/              # Agent definitions
├── bin/
│   └── install.js       # npm installer script
├── landing-page/        # Documentation website
├── migrations/          # Migration system
├── AGENTS.md            # Agent dev guidelines
├── AGENT-WORKSPACE.md   # Workspace template
├── ai-function.sh       # Shell function definition
├── install.sh           # Installer script
├── package.json         # npm package config
└── README.md            # This file

Usage

AI CLI

ai init        # Initialize directory with AGENTS.md and AGENT-WORKSPACE.md
ai update      # Update local repo with latest changes
ai update-all  # Update all registered directories
ai list        # List all registered directories
ai forget      # Remove current directory from registry

How ai init Works

1. Clone/update repo: Clones or updates ~/.ai-repo-local-clone (shallow clone)
2. Create symlinks: Creates symlinks in current directory:
   • AGENTS.md → ~/.ai-repo-local-clone/AGENTS.md
   • CLAUDE.md → ~/.ai-repo-local-clone/CLAUDE.md
3. Copy template: Copies AGENT-WORKSPACE.md template (if not exists)
4. Register directory: Adds current directory to ~/.ai-registry for bulk updates

This means your AGENTS.md and CLAUDE.md always stay up-to-date via symlinks. Run ai update to refresh the local repo.

Registry System

The registry tracks all initialized directories to enable bulk updates:

• ai list - Shows all registered directories (marks missing ones with ✗)
• ai update-all - Updates all registered directories in one command
• ai forget - Removes current directory from registry (doesn't delete files)

Registry location: ~/.ai-registry (plain text, one path per line)

For AI Agents

Read in this order:

1. AGENTS.md - Universal best practices
2. AGENT-WORKSPACE.md - Workspace-specific context

For Developers

Consult AGENTS.md when:

• Building AI agent systems
• Making architectural decisions (create ADR first)
• Establishing code standards for new projects
• Training agents on production practices
• Reviewing code quality

Use AGENT-WORKSPACE.md to:

• Document project-specific patterns
• Track current work and context
• Define development commands
• Provide workspace setup instructions

Key Principles

• Explicit over implicit
• Fail fast, fail loud
• Optimize for deletion
• Test-first development (non-negotiable)
• ADRs required for architectural decisions
• Zero tolerance for weak types, magic values, commented code