Agentic Os

Turns Claude Code into a persistent operating system with a kernel, specialist agents, and file-based memory. You write a CLAUDE.md kernel that routes tasks to agents like @dev, @writer, or @researcher, each defined as markdown files with their own tools and memory scope. Slash commands become reusable workflows, daily logs replace databases, and everything survives session restarts because it's just files in your project. The architecture is lifted from production agentic systems but stripped down to markdown tables and JSON instead of infrastructure. If you're building long-running projects where context matters or want Claude to run morning briefings and scheduled tasks, this is the blueprint.