Model or dataset
elirantutia/vibeyard avatar
elirantutia/vibeyard

Vibeyard: an Electron IDE that wraps Claude Code, Codex CLI and Gemini CLI sessions

The IDE built for AI coding agents.

1,385 stars173 forksTypeScriptMIT

At a glance

What is it?
Vibeyard is a terminal-centric desktop IDE for running AI coding agent CLIs. It adds a per-project dashboard, a kanban board, parallel PTY sessions, cost tracking and P2P session sharing on top of tools you already have installed.
Who is it for?
Adopt Vibeyard if you already run Claude Code, Codex CLI or Gemini CLI in a terminal and the tab juggling has become the bottleneck: the kanban board, per-session cost readout and resume behaviour are the parts worth the install. Skip it if you want a general-purpose editor, since the project describes itself as terminal-centric and its value sits in the session layer, not in language tooling.
Can I use it commercially?
Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 28 days ago.
What is it written in?
Mainly TypeScript, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on October 1, 2026, and from our analysis. They are not legal advice.

Editorial analysis

The terminal juggling problem Vibeyard targets

Running several AI coding agents at once in a plain terminal produces a specific kind of mess. Each agent needs its own PTY, each one holds a context window you cannot see, and each one bills you by the token. When you switch projects you either kill sessions or leave them running in tabs you have lost track of. Vibeyard's README frames the project as a response to exactly that: "Running AI coding agents in a bare terminal gets messy fast." The audience is narrow and identifiable. You must already have at least one supported CLI installed and authenticated, meaning Claude Code, OpenAI Codex CLI or Gemini CLI. Vibeyard does not provide the model access itself; it is a workspace around command line agents that exist independently. If you do not use any of those three tools, the app has nothing to wrap.

How the session layer is put together

The package description calls Vibeyard "a terminal-centric IDE for AI-powered CLI tools, built on Electron", and the repository layout matches that claim: separate TypeScript configs for the main process, the preload script, the renderer and the tests, plus an esbuild step that bundles src/renderer/index.ts into a browser-targeted IIFE. The xterm topic in the repository metadata points at the terminal emulation layer, and the README states that each agent session runs in its own PTY. Everything above that is orchestration. A per-project dashboard holds draggable widgets; the kanban board stores cards that can spawn or resume a CLI session; a card moves to Done when its session completes. The session inspector exposes telemetry for a running session: timeline, cost breakdown, tool usage stats and context window monitoring. Multiple Claude profiles are handled by giving each session its own isolated config directory, so credentials, settings and history stay separate, and you can pick a profile per session or set a default per project or globally. P2P sharing uses WebRTC with read-only or read-write modes and PIN authentication. That is the whole architecture as documented: Electron shell, PTYs, per-session config directories, and a renderer that renders dashboards and terminals.

Installing Vibeyard and starting a first session

The prerequisite comes before any install command: at least one supported CLI must be installed and authenticated. After that, the fastest path on macOS, Linux or Windows is the npm wrapper. The package ships a bin entry named vibeyard, and the README says the app is downloaded and launched automatically on first run.

bash
npm i -g vibeyard
vibeyard

If you prefer a native artifact, the README points to GitHub Releases and lists a .dmg for macOS (signed and notarized by Apple), .deb and .AppImage for Linux, and a Setup .exe plus a portable .exe for Windows. The Linux commands are given directly in the README:

bash
sudo dpkg -i vibeyard_*.deb
chmod +x Vibeyard-*.AppImage
./Vibeyard-*.AppImage

For a source build, the README gives a three-step sequence. Note that it states Node v24+ and points at .nvmrc, while package.json declares engines.node as >=18; the README is the stricter of the two and is the one to follow for a working build.

bash
git clone https://github.com/elirantutia/vibeyard.git
cd vibeyard
npm install && npm start

Once the app is open, the first real use is to create or open a project, then start an agent session in it. The README documents Cmd+\ as the shortcut for spinning up a new session, and Cmd+Shift+I for the session inspector. Swarm mode arranges all sessions for a project in a grid, which is the view to open when you are running more than two agents at once. If you have both a work and a personal Claude Code login, create a second Claude profile before starting sessions, because profile selection happens per session and the isolation is at the config directory level.

Where Vibeyard stops being the right tool

Vibeyard is not a code editor in the sense that VS Code or JetBrains IDEs are. The README does not describe language servers, refactoring tools, debuggers or a plugin marketplace, and the embedded browser tab exists to send DOM selectors and page URLs to an agent rather than to serve as a general web development surface. If your work is mostly writing code by hand with an agent used occasionally, you are paying the cost of a second application for features you will not use. The provider list is also a hard boundary: three CLIs are supported, and the README says more are coming, which is a statement about intent rather than a shipped capability. Cost and context tracking can only report what the underlying CLI exposes, so their accuracy is bounded by those tools. The README does not document rollback, migration between versions, or what happens to stored session state when you upgrade, so treat session history as something to verify on your own machine before you depend on it. Finally, P2P sharing over WebRTC with PIN authentication is a convenience feature for teammates, not an access control system, and the README does not describe audit logging.

How it compares with tmux plus a terminal multiplexer setup

The honest alternative for many people is not another GUI IDE. It is tmux, or a terminal emulator with split panes, plus the agent CLIs run directly. That setup costs nothing, has no Electron process, and works over SSH on a remote box, which Vibeyard does not claim to do. The difference in approach is where state lives. With tmux, the session is the process and your mental model of it is the pane title you typed. Vibeyard moves that state into an application: cards on a kanban board, cost and token counters per session, colour-coded status dots on tabs, a resume feature that survives restarting the app, and desktop notifications when a session needs input. You get that bookkeeping only if the app stays running and its stored state stays intact. A second alternative is staying inside a full IDE that has already added agent integration, which keeps your editor and your agent in one window but gives you less control over running several providers side by side in isolated config directories. Vibeyard's bet is that parallel sessions across multiple providers is the workflow worth building a separate app for.

Maintenance, releases and what the MIT licence means here

The repository is not archived, and the last push was on 2026-09-03, the same day v0.3.8 was published. The release cadence visible in the release list is tight: v0.3.6 on 2026-08-24, v0.3.7 on 2026-08-31, v0.3.8 on 2026-09-03. That is a project shipping on a roughly weekly rhythm, and the version numbers are still in the 0.x range, which is a fair signal that interfaces and stored state can change between releases. Upgrade cost is therefore real: an Electron app that tracks sessions, costs and per-profile config directories has state to migrate, and the README does not describe a migration path. Budget for reading CHANGELOG.md before each upgrade. On licensing, Vibeyard is MIT, which permits commercial use, modification and redistribution provided the copyright notice and permission notice are retained. One thing the MIT grant does not cover is the CLIs it drives: Claude Code, Codex CLI and Gemini CLI are separate products with their own terms, and the README itself carries a line stating that Vibeyard is an independent project and is not affiliated with or endorsed by Anthropic. Anyone evaluating this for a company should check the terms of the agent CLIs separately, and treat that as a question for counsel rather than something the repository answers.

Editorial conclusion

Adopt Vibeyard if you already run Claude Code, Codex CLI or Gemini CLI in a terminal and the tab juggling has become the bottleneck: the kanban board, per-session cost readout and resume behaviour are the parts worth the install. Skip it if you want a general-purpose editor, since the project describes itself as terminal-centric and its value sits in the session layer, not in language tooling. Before committing, install a supported CLI and authenticate it, then check that the desktop build you download matches your platform, because the README lists separate macOS, Linux and Windows artifacts and an npm wrapper that downloads the app on first run.

Frequently asked questions

What is Vibeyard used for?

It is a desktop IDE for managing AI coding agent sessions. The README describes a per-project dashboard, a kanban task board, parallel sessions each in their own PTY, cost and context tracking, and session resume, all wrapping Claude Code, OpenAI Codex CLI or Gemini CLI.

Which AI CLIs does Vibeyard support?

The README names Claude Code, OpenAI Codex CLI and Gemini CLI, and says more providers are coming. At least one of them must be installed and authenticated before Vibeyard is useful.

How do I install Vibeyard on Linux?

Download the .deb or .AppImage from GitHub Releases. The README gives sudo dpkg -i vibeyard_*.deb for Debian and Ubuntu, or chmod +x followed by running the AppImage for the universal build. There is also an npm wrapper, npm i -g vibeyard, which downloads and launches the app on first run.

Can Vibeyard run multiple Claude Code logins at the same time?

Yes. The README describes multiple Claude profiles where each session is backed by its own isolated config directory so credentials, settings and history do not mix. You pick a profile per session, or set a per-project or global default.

Is Vibeyard affiliated with Anthropic?

No. The README states that Vibeyard is an independent project and is not affiliated with or endorsed by Anthropic.

Official sources

  1. elirantutia/vibeyard on GitHub
  2. Issues
  3. License: MIT
  4. README
  5. Releases
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/elirantutia-vibeyard.svg)](https://hysenlabs.com/projects/elirantutia-vibeyard)