create-pga-ai

The fastest way to start building with PGA Platform — Create a complete PGA agent project with a single command.

npm create pga-ai@latest my-agent

That's it! 🎉

What You Get

✨ Interactive Setup — Answer a few questions, get a fully configured project 🧬 Evolution Boost 2.0 — 10x faster agent evolution built-in 🔌 Multiple LLM Support — Anthropic Claude, OpenAI GPT, or both 💾 Production-Ready Storage — PostgreSQL or in-memory for development 📦 Zero Configuration — Everything works out of the box 🚀 Template System — Pre-built agents for common use cases

Quick Start

1. Create Your Project

npm create pga-ai@latest my-agent
cd my-agent

2. Configure API Keys

# Edit .env with your API keys
nano .env

3. Run Your Agent

npm run dev

That's it! Your agent is now running with PGA's self-evolution capabilities.

Usage

Interactive Mode (Recommended)

Simply run without arguments for an interactive setup:

npm create pga-ai@latest

You'll be asked:

• Project name — Your agent's folder name
• LLM Provider — Anthropic Claude (recommended), OpenAI GPT, or both
• Storage — PostgreSQL (production) or in-memory (development)
• Evolution Boost — Enable 10x faster evolution (recommended: Yes)
• Template — Choose a starter template

Command-Line Mode

Skip the prompts with CLI flags:

npm create pga-ai@latest my-agent \
  --template chatbot \
  --llm anthropic \
  --storage postgres \
  --boost

Available Options

| Option | Values | Description | |--------|--------|-------------| | --template | chatbot, code-assistant, customer-support, data-analysis, custom | Project template | | --llm | anthropic, openai, both | LLM provider | | --storage | postgres, memory | Storage backend | | --boost | flag | Enable Evolution Boost 2.0 | | --skip-install | flag | Skip npm install |

Templates

🤖 Chatbot Agent

General-purpose conversational agent. Perfect for:

• Customer service bots
• Personal assistants
• Interactive Q&A systems

💻 Code Assistant

Programming helper with code understanding. Great for:

• Code review automation
• Documentation generation
• Developer productivity tools

🎧 Customer Support

Specialized support automation. Ideal for:

• Ticket triage
• FAQ automation
• Support escalation

📊 Data Analysis

Analytics and insights expert. Perfect for:

• Business intelligence
• Data visualization
• Report generation

⚙️ Custom

Blank starter for your own use case. Maximum flexibility.

Project Structure

my-agent/
├── src/
│   ├── index.ts        # Main entry point
│   └── agent.ts        # Agent logic
├── .env                # API keys & configuration
├── .env.example        # Example environment file
├── .gitignore          # Git ignore rules
├── package.json        # Dependencies & scripts
├── tsconfig.json       # TypeScript configuration
└── README.md           # Project documentation

Evolution Boost 2.0

All projects include Evolution Boost 2.0 — the world's first genetic algorithm for prompt evolution.

Regular Evolution:

• 20-30 generations for 2x improvement
• Sequential mutations
• 8-15% improvement per generation

Evolution Boost 2.0:

• 2-3 generations for 2x improvement (10x faster!)
• Parallel mutations (10 branches simultaneously)
• 40-80% improvement per generation
• Meta-learning system
• Pareto optimization

Enable it during setup or in your .env:

EVOLUTION_MODE=aggressive  # Options: conservative, balanced, aggressive

What Gets Installed

Depending on your configuration, the installer adds:

Core Package:

• @pga-ai/core — Main PGA framework

LLM Adapters:

• @pga-ai/adapters-llm-anthropic — Claude integration
• @pga-ai/adapters-llm-openai — GPT integration

Storage Adapters:

• @pga-ai/adapters-storage-postgres — PostgreSQL storage

Dev Dependencies:

• typescript — TypeScript compiler
• tsx — TypeScript execution
• vitest — Testing framework
• @types/node — Node.js types

Requirements

• Node.js — v18 or higher
• npm — v9 or higher
• PostgreSQL — (if using postgres storage)

Examples

Create a Chatbot with Claude

npm create pga-ai@latest my-chatbot \
  --template chatbot \
  --llm anthropic \
  --boost

Create a Code Assistant with Both LLMs

npm create pga-ai@latest code-helper \
  --template code-assistant \
  --llm both \
  --storage postgres \
  --boost

Create Custom Agent (Development)

npm create pga-ai@latest my-agent \
  --template custom \
  --llm openai \
  --storage memory

Development Scripts

Generated projects include these npm scripts:

npm run dev      # Run with hot-reload (tsx)
npm run build    # Compile TypeScript
npm start        # Run compiled version
npm test         # Run tests

Environment Variables

The installer creates a .env file with these variables:

# Anthropic API (if selected)
ANTHROPIC_API_KEY=your-api-key-here

# OpenAI API (if selected)
OPENAI_API_KEY=your-api-key-here

# PostgreSQL (if selected)
DATABASE_URL=postgresql://user:password@localhost:5432/pga_db

# Evolution Configuration
EVOLUTION_MODE=balanced  # conservative | balanced | aggressive

Troubleshooting

"Permission denied" error

Make sure you're using npm 7+ which supports npm create:

npm --version  # Should be 7.0.0 or higher
npm update -g npm

"Module not found" errors

Ensure dependencies installed correctly:

cd my-agent
rm -rf node_modules package-lock.json
npm install

TypeScript errors

Check your Node.js version:

node --version  # Should be v18 or higher

Learn More

• Documentation: https://pga.ai/docs
• API Reference: https://pga.ai/api
• Discord Community: https://discord.gg/pga
• GitHub: https://github.com/pga-platform

License

MIT © PGA Platform

Built with PGA 🧬 — The world's first genomic self-evolving AI system