One window for a whole swarm of AI coding agents — split terminals, per-agent git worktrees, live web previews, and a War Room where agents collaborate
# Add to your Claude Code skills
git clone https://github.com/duongducnguyen/swarmtermswarmterm is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by duongducnguyen. One window for a whole swarm of AI coding agents — split terminals, per-agent git worktrees, live web previews, and a War Room where agents collaborate. It has 50 GitHub stars.
swarmterm's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/duongducnguyen/swarmterm" and add it to your Claude Code skills directory (see the Installation section above).
swarmterm is primarily written in TypeScript. It is open-source under duongducnguyen on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh swarmterm against similar tools.
No comments yet. Be the first to share your thoughts!
Unlocks once the catalog security scan passes (runs nightly).
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
One window for a whole swarm of AI coding agents.
Real split terminals, one git worktree per agent — and an app your agents can drive themselves.
Running one coding agent in a terminal works anywhere. Running five is where ordinary terminals fall apart: tabs hide who is doing what, agents trample each other's edits, and you become the relay pasting prompts and URLs between windows.
Swarmterm is a native desktop app (Tauri 2 + Rust, real PTY terminals, a VS Code-flavoured interface) built around three ideas:
See the whole swarm. A workspace is one window of real split terminal panes — live titles, an activity dot on any pane still producing output, and broadcast typing for when five agents should hear the same thing at once.
Keep the work separate. Flip one switch and every agent pane is born inside its own git worktree, on its own branch — separate, reviewable diffs from the very first edit, with a git panel to read them without leaving the app.
Agents are users of the app, too. Every pane is connected to Swarmterm itself, so agents don't just run in it: they put a live web preview beside their own pane, spin up isolated worktrees to delegate subtasks, and discuss work with each other — with you moderating.
Everything else hangs off those three ideas. The full tour — panes, broadcast, session resume, the web preview, the War Room, the status line, shortcuts, troubleshooting — lives in the user guide.
Download from the latest release:
Swarmterm_x.y.z_x64-setup.exe. Not code-signed yet, so SmartScreen warns on first run: More info → Run anyway.Swarmterm_x.y.z_universal.dmg, signed and notarized (Intel + Apple Silicon).Both update themselves in-app from then on. Install whichever agent CLIs you plan to use (claude, codex, opencode) — anything missing just appears disabled.
The route for Linux (no packaged build yet) and for hacking on Swarmterm:
git clone https://github.com/duongducnguyen/swarmterm.git
cd swarmterm
npm install
npm run tauri dev # develop / just use it
npm run tauri build # installer / app bundle
You need Node.js 18+ (npm 10+) and a stable Rust toolchain from rustup.rs, plus per platform: on Windows, the WebView2 Runtime (usually preinstalled) and Microsoft C++ Build Tools; on macOS, Xcode Command Line Tools (xcode-select --install); on Linux, webkit2gtk-4.1, librsvg, libayatana-appindicator3 and build-essential — see Tauri prerequisites. The first build compiles the Rust side and takes a few minutes.
Swarmterm opens on the composer: pick a folder, choose how many terminals, assign Claude Code, Codex or OpenCode to some of them, switch on Isolate features in git worktrees, and press ⌘/Ctrl + Enter. Your panes open with the agents already starting up; the right panel gives you the web preview, the git view, and the War Room.
Two things worth knowing up front: nothing is saved between launches (a clean slate every start is deliberate — Settings do persist), and closing the window hides the app to the tray with everything still running; tray → Quit is what actually exits.
From here, the user guide covers the rest.
1.0 is out — installers for macOS (signed & notarized) and Windows, with in-app auto-update — and it moves fast as a daily driver for parallel agent work. Honest reasons to wait:
The full list lives in the guide's known limits.
Bug reports and questions are welcome in issues; CONTRIBUTING.md covers setup, the test/CI checklist, and how to propose features — design discussion happens in an issue first for anything non-trivial. Swarmterm is free software under GPL-3.0. Official release builds send an anonymous launch ping (app version + OS, nothing else) so we can see how many people use Swarmterm; builds from source contain none — details in the guide's telemetry section.
Built on Tauri, xterm.js and portable-pty; interface styled after VS Code. Working on Swarmterm itself? Start with CONTRIBUTING.md, CLAUDE.md, and the release checklist in docs/manual-smoke-tests.md.