Hey Hunters 👋 I built Cockpit because using Claude Code at scale was killing my flow. I'd have 4 windows open: chat in one, editor in another, terminal in a third, and a browser tab to verify results. Switching between them ate more time than the actual coding. So I asked: pilots don't fly with 5 separate dashboards — why should we? Cockpit is a unified development cockpit built on the Claude Code Agent SDK. Three panels, swipe-to-switch, always rendered: • Agent — multi-project AI sessions in parallel, red-dot notifications when done • Explorer — file tree, syntax highlighting (Shiki), Git blame/diff, LSP, fuzzy search • Console — terminal + smart "bubbles" for browser automation, Postgres/MySQL/Redis The killer feature is parallel sessions: kick off 5 Claude tasks across different projects, switch between them while they run, get notified when each one completes. I went from running 1 AI task at a time to ~10. Zero cloud dependency. If you can run Claude Code, Cockpit works — same API key, no signup, all local. 🚀 Try without installing: https://cocking.cc/try (5-minute sandbox) 📦 GitHub: https://github.com/Surething-io/... ⌨️ Install: npm i -g @surething/cockpit Would love feedback — especially on the multi-session UX. Building this changed how I work; hoping it does for you too.

Cockpit

The local-first cockpit for AI coding at scale

The local-first cockpit for AI coding at scale