P
Personal AI Infrastructure
Open-source scaffolding for building your own AI-powered operating system
The best AI in the world should be available to everyone
The Evolution: From Mirror Platform to PAI Packages
PAI 1.0.0 represents a fundamental shift: from a monolithic "mirror my exact system" approach to a modular, package-based architecture that democratizes AI infrastructure.
⚠️
The Problem
The Jenga Tower Effect
Early PAI was a monolithic system where everything depended on everything else. Want to use just one skill? Too bad—you had to clone the entire infrastructure.
Updates broke workflows. Dependencies tangled. The system became fragile, like a Jenga tower where pulling one block could collapse everything.
The Solution
Modular Packages
PAI 1.0.0 introduces self-contained packages. Each package is a complete unit with its own dependencies, documentation, and installation scripts.
Install only what you need. Update packages independently. Build your own packages. Share them with the community. The infrastructure serves you—not the other way around.
What This Means for You
Democratized Contribution
Anyone can create and share packages without understanding the entire PAI architecture.
Platform Agnostic
Packages work across any AI platform. Your skills, your agents, your infrastructure.
Reduced Maintenance
Each package is self-contained. Update one without breaking others.
Fabric Integration
248 Fabric patterns run natively in your context. No CLI spawning required.
The Four Primitives
PAI is built on four core primitives that work together to create a powerful, flexible AI infrastructure tailored to your needs.
Skills
Core FoundationModular capabilities you can install and extend
- ✓50+ built-in skills
- ✓Custom skill creation
- ✓MCP server integration
- ✓Self-contained documentation
Agents
Specialized AI personas with unique expertise
- ✓Parallel delegation
- ✓Custom personalities
- ✓Model selection per agent
- ✓Context inheritance
Hooks
Event-driven automation and workflows
- ✓Session lifecycle hooks
- ✓Skill-level hooks
- ✓System event triggers
- ✓Async execution
History
Unified Observation and Context System (UOCS)
- ✓Automatic capture
- ✓Searchable archive
- ✓Persistent learning
- ✓Privacy-first storage
The 13 Founding Principles
These principles define how to build reliable, scalable, and maintainable AI infrastructure. Click any principle to explore it in depth.
PRINCIPLE 01
Clear Thinking + Prompting is King
The quality of outcomes depends on the quality of thinking and prompts. Clear thinking comes before code.
Learn More
PRINCIPLE 02
Scaffolding > Model
System architecture matters more than the underlying AI model. Structure outperforms raw power.
Learn More
PRINCIPLE 03
As Deterministic as Possible
Favor predictable, repeatable outcomes over flexibility. Same input → Same output. Always.
Learn More
PRINCIPLE 04
Code Before Prompts
Solve problems with code first. Only use prompts when code cannot handle the task.
Learn More
PRINCIPLE 05
Spec / Test / Evals First
Define specifications, write tests, and create evaluations before implementation.
Learn More
PRINCIPLE 06
UNIX Philosophy
Do one thing well. Compose small, focused tools into powerful systems.
Learn More
PRINCIPLE 07
ENG / SRE Principles
Build for reliability, observability, and operability from day one.
Learn More
PRINCIPLE 08
CLI as Interface
Command-line interfaces are universal, scriptable, and composable.
Learn More
PRINCIPLE 09
Goal → Code → CLI → Prompts → Agents
Build from concrete to abstract: goals become code, code exposes CLIs, CLIs enable prompts, prompts power agents.
Learn More
PRINCIPLE 10
Meta / Self Update System
The system should improve itself. Capture learnings and evolve automatically.
Learn More
PRINCIPLE 11
Custom Skill Management
Skills are the unit of capability. Manage them like code packages.
Learn More
PRINCIPLE 12
Custom History System
Context is everything. Automatically capture, organize, and surface relevant history.
Learn More
PRINCIPLE 13
Custom Agent Personalities
Different tasks need different personas. Customize agent behavior and expertise.
Learn More
How It Works
PAI is built on a layered architecture where each component works together to create a powerful, extensible AI infrastructure.
1
Skill Routing
Intelligent routing system directs requests to the appropriate skill based on patterns and keywords.
2
Agent Delegation
Parallel agent execution with context inheritance and specialized model selection per agent.
3
Hook Events
Lifecycle hooks capture session start/end, skill execution, and system events for automation.
4
History Capture
UOCS automatically documents all work with structured metadata for searchable context.
Get Started
Install PAI in minutes. Choose your platform and follow the steps below.
1
Clone PAI
git clone https://github.com/danielmiessler/PAI.git ~/PAI2
Create Symlink
[ -d ~/.claude ] && mv ~/.claude ~/.claude.backup
ln -s ~/PAI/.claude ~/.claude3
Run Setup Wizard
~/.claude/Tools/setup/bootstrap.sh4
Add Your API Keys
cp ~/.claude/.env.example ~/.claude/.env
nano ~/.claude/.env5
Start Claude Code
source ~/.zshrc # Load PAI environment
claudeNote: The setup wizard will configure your name, email, AI assistant name, and environment variables to customize PAI to your environment.
Join the Community
PAI is built by and for the community. Connect with other builders, share your packages, and help shape the future of personal AI infrastructure.
Ready to Build Your AI Infrastructure?
The best AI in the world should be available to everyone. Join us in building the future of personal AI systems.