The Problem
AI-assisted development works better when the environment is designed for it — not assembled ad-hoc. Context gets lost between sessions. There's no standard way to persist knowledge, enforce project conventions, or surface system state while deep in a coding session.
The Approach
Custom Arch Linux built from scratch around Claude Code as the primary development tool. The work isn't the OS itself — it's the layer on top: a structured memory system with its own file format and retrieval conventions, shell hooks that enforce consistency and automate bookkeeping, CLAUDE.md scaffold files that give Claude Code project-specific context, and a Wayland HUD that keeps system state visible without alt-tabbing. The environment travels on a USB drive.
The Result
A development environment that accumulates knowledge across sessions. The memory system persists decisions, gotchas, and patterns so the next session starts with full context. Scaffold architecture means Claude Code produces structurally sound output — the constraints are built into the project layout, not re-explained every prompt. VibeArch is the methodology made tangible: every project on this site was developed inside it.
Architecture
Demo & Screenshots
Tech Stack
Rolling release, minimal base, total control over every package. No bloat, no assumptions about what a development environment should look like.
Primary development tool. The entire environment is engineered to maximize its effectiveness: scaffold architecture, persistent memory, hook-enforced conventions.
Hooks, automation, system glue. The memory system, validation logic, and session management are all shell scripts — readable, portable, no dependencies.
Modern display protocol via Hyprland compositor. Tiling window management, custom HUD integration, clean multi-monitor handling.
Current Status
Daily driver. This is the system everything else is built on — every project on this site was developed inside VibeArch.
What's Next
HUD framework rebuild with modular plugin architecture, expanded local inference model support, improved system monitoring.