PAI — Personal AI Infrastructure

A framework for building a persistent, skill-based AI assistant on top of Claude Code.

What This Is

PAI turns Claude Code into a personalized AI system with:

• The Algorithm — A structured 7-phase execution framework (Observe → Think → Plan → Build → Execute → Verify → Learn) that produces verifiable outcomes using Ideal State Criteria
• Skills — Modular capability packages with workflows, tools, and routing
• Hooks — Lifecycle event handlers for security validation, context management, session learning, and system integrity
• Memory — Persistent file-based memory system for user context, project state, and learning reflections
• TELOS — A life operating system for tracking goals, projects, challenges, and personal growth

Origin

This framework was originally designed and built by Daniel, who open-sourced it. This fork represents an evolution tailored to a specific user's mission — systematic trading automation, Linux cloud infrastructure (Fedora Cloud SIG), and bridging the two.

We're sharing it back because the architecture is genuinely useful and because building on someone else's thoughtful work deserves acknowledgment.

Structure

~/.claude/
├── PAI/                    # Core system (Algorithm, context routing, tools)
│   └── Algorithm/          # Versioned execution framework
├── skills/                 # Modular capabilities
│   ├── Trading/            # Market analysis, game plans, trade review
│   ├── Research/           # Multi-agent parallel research
│   ├── Thinking/           # FirstPrinciples, Council, RedTeam, Science
│   ├── Telos/              # Life OS — goals, projects, learning
│   ├── Agents/             # Custom agent composition
│   ├── USMetrics/          # Economic indicators
│   └── Utilities/          # System tools (upgrade, scaffolding, delegation)
├── hooks/                  # Lifecycle event handlers
├── MEMORY/                 # Persistent state (work, reflections, snapshots)
└── settings.json           # Configuration (permissions, hooks, MCP)

Getting Started

1. Install Claude Code
2. Copy this framework to ~/.claude/
3. Create your PAI/USER/ directory with your own TELOS, projects, and customizations
4. Create a settings.json from settings.json.template

License

MIT — Use it, fork it, make it yours.

Acknowledgments

• Daniel — Original PAI architect who open-sourced this system. Thank you for designing something worth building on.
• Anthropic — For Claude Code and the model that makes this possible.