Over the past year I’ve settled into a terminal-first setup that I’m genuinely happy with. Here’s a quick overview of how the pieces fit together.

tmux as the foundation

tmux is the backbone of everything. I run all my work inside persistent sessions — one session per project. Vim-style pane navigation (h/j/k/l) keeps my hands off the mouse, and vi mode in copy mode makes scrolling back through output feel natural.

Neovim

My Neovim config is built on top of kickstart.nvim. It’s a minimal, well-documented starting point that I can actually understand and maintain. LSP, autocompletion, and fuzzy finding out of the box without a ton of magic.

Obsidian for notes

I keep a local Obsidian vault for notes, journaling, and thinking through problems. It lives alongside my code projects so the sessionizer can drop me into it just as easily as any repo. Having notes in the same workflow (rather than a separate app) means I actually use them.

Claude Code in the loop

Claude Code runs directly in the terminal inside tmux, which means it fits naturally into the existing setup. No browser tab switching, no separate IDE. I can have Claude working in one pane while I review the output in another, and my Obsidian notes are one keypress away if I need to pull up context or jot something down.

The combination of fast project switching, persistent sessions, and an AI assistant in the terminal has made context switching much less painful. Each piece is simple on its own — the value comes from how they connect.