core_skill

Overview

This skill is the authoritative core for Personal AI Infrastructure (PAI). It auto-loads at session start and defines how the system behaves, how to configure it, and how to use system-level workflows securely and predictably. It exists to make agentic assistance reliable, auditable, and focused on amplifying human capability.

How this skill works

CORE initializes on session start and exposes the canonical architecture, principles, and configuration patterns for PAI. It documents lifecycle hooks, memory handling, agent types, security boundaries, notification routing, and workflow triggers so agents and users behave consistently. CORE enforces validation-first workflows: changes must be made, tested, and verified before reporting success.

When to use it

• At the start of every session to load system context
• When you need a canonical answer about system identity, configuration, or security
• Before creating or customizing agents or workflows
• When auditing system integrity, secrets, or repository boundaries
• When routing work through background agents or delegation patterns

Best practices

• Always consult CORE before changing system-level settings or workflows
• Use USER-level overrides for private customizations; CORE defaults remain safe
• Validate fixes with automated tests and visual verification before claiming completion
• Keep sensitive data in the private local instance and never expose it publicly
• Follow the two-tier SYSTEM/USER pattern so updates don’t overwrite personal configs

Example use cases

• Explain how agent types differ and when to invoke a named agent versus a task subagent
• Walk through the memory capture model so session insights compound over time
• Run a secret-scan and integrity audit before pushing commits
• Configure notification routing for voice, push, and team channels
• Design a delegation plan that spawns parallel agents for a complex project