What is nodeterm?
nodeterm turns your terminals and coding agents into live nodes on an infinite canvas. Instead of stacking sessions into tabs you’ll forget about, you lay them out spatially — drag them, group them, zoom out to see everything that’s running, zoom in to work.
Every node is a real terminal. Agent nodes run the real CLIs — Claude Code, Codex, Gemini, opencode — and the canvas shows you what they’re doing: a pulsing RUNNING badge while an agent works, NEEDS YOU when it’s waiting on input, live subagent fan-outs, and a notification when a long task finishes while you’re elsewhere.
Why a canvas?
Section titled “Why a canvas?”Tabs hide work; a canvas shows it. nodeterm is built for people who run many things at once — several agents on several projects — and think spatially about what’s where. If you’ve ever lost a running task behind a stack of tabs, this is for you.
Sessions that don’t die
Section titled “Sessions that don’t die”Terminals run inside tmux sessions that outlive the app. Quit nodeterm, relaunch it, and every terminal is exactly where you left it — running processes included. Even after a machine reboot, nodeterm restores each terminal’s scrollback and relaunches your agent sessions where they left off.
More than terminals
Section titled “More than terminals”The canvas also holds sticky notes, group frames, a Monaco code editor, git diff views, and SDK-driven Claude chat nodes — and agents can be wired together so one can read another’s context.
Where it runs
Section titled “Where it runs”- macOS & Linux desktop app — the primary experience (Apple Silicon and Intel on
macOS; x64 AppImage or
.debon Linux). - Server Edition — the same canvas served from a headless Linux box to your browser.
- iOS companion — attach to your sessions from your phone.