Session Monitoring

Know what every AI coding agent is doing — from your Mac notch

You don't run one AI coding agent anymore. You run four. One is refactoring a module in a terminal tab you can't see, another is waiting on a permission prompt you forgot about, a third finished ten minutes ago and has been sitting idle ever since. The real cost of all this parallelism isn't compute — it's attention. Vibe Island puts the live status of every AI coding session into your Mac's notch, so you always know what each agent is doing without hunting through windows.

Download Free Trial macOS 14+ · 2-day free trial · no card required
Vibe Island FileEditWindowHelp
fix auth bug 3
fix auth bug ClaudeiTerm 27m
You: fix the auth bug in middleware
Writing middleware.ts
backend server
CodexTerminal1h
optimize queries
GeminiGhostty5h
Permission Request
Edit src/auth/middleware.ts
12const verify = (token) =>
13-  jwt.verify(token);
13+  if (!token) throw new
14+   AuthError('missing');
15+  return jwt.verify(token,
+3 -1
Claude asks
Which deployment target?
fix auth bug ClaudeiTerm 28m
You: fix the auth bug in middleware
Done — click to jump
backend server
CodexTerminal1h
optimize queries
GeminiGhostty5h
Approved
gemini — optimize-queries
Analyzing the slow queries.
Read(schema.prisma)
└── 1.2 KB
Edit(src/db/queries.ts)
└── Updated (+8 -23)
codex — backend-server
Building the REST endpoints.
Write(src/routes/users.ts)
└── New file (47 lines)
Bash(npm test)
└── 3 passed
claude — fix-auth-bug
Let me look at the auth module.
Searching for 6 patterns… (ctrl+o to expand)
Read 2 files (ctrl+o to expand)
 
Found the issue — token validation skips expiry check.
Terminal
Vibe Island
Every agent. One glance.
Claude Code, Codex, Gemini CLI, and Cursor — all in a single view.

Every session in one place

Each AI coding agent you have running shows up in a single notch panel with a live, color-coded status: working, waiting on you, or done. No more cycling through terminal tabs to find out which agent stalled and which one quietly finished. The panel updates the moment a session changes state, so a glance at the top of your screen tells you everything — which agents are busy, which are blocked, and which are ready for your next move. When nothing needs you, the notch stays quiet; the information is there when you want it and invisible when you don't.

The moment it needs you, you know

The whole point of an agent is that it works while you do something else. The catch is that it occasionally needs you — to approve a risky action, answer a question, or confirm a plan before it continues. Vibe Island surfaces that moment the instant it happens. You can approve or deny a permission request straight from the notch, answer a multiple-choice question with a single click, and review a plan with its full formatting before you let an agent proceed — all without switching windows or breaking your focus. You get pulled back exactly when it matters, and left alone when it doesn't. That is the difference between an agent that interrupts you and one you stay quietly in control of.

Usage and rate limits, tracked too

Running this many agents burns through subscription quota fast. Vibe Island tracks each provider's remaining limit and window reset right beside the session status, and brings full token and cost breakdowns — input, output, cache, and per-model spend, by day and by month — into the same panel. You see how close you are to a rate limit before it stops you mid-task, with no separate usage app to keep open. See the full picture on the usage tracking page.

Run many sessions in parallel

Watching one agent is easy. Watching 25 at once is where most setups fall apart. Vibe Island gives every concurrent session its own brand-colored tag so you can tell them apart at a glance, and lets you switch between them with one click. Whether you've got two agents on one project or a dozen spread across repos, they all live in the same notch. Read more about running multiple agents in parallel.

Jump straight to any session

Knowing a session needs you is only half the job — you still have to get to it. Click any session in the notch and Vibe Island lands you on the exact terminal tab, split pane, or editor window where that agent is running, across 18+ terminals and editors. No guessing which window, no scrubbing through tabs. The session you clicked is the one in front of you. See how precise terminal jump works.

Track remote sessions over SSH

Plenty of agents don't run on your laptop at all — they run on a remote box with the real horsepower. Vibe Island monitors agents running on remote servers over SSH and shows them in the same notch as your local ones, with the same live status and alerts. It reconnects on its own when the tunnel drops, so a flaky connection never leaves you blind. Learn more about monitoring remote agents over SSH.

Works with 25+ AI coding agents

Session monitoring isn't tied to one tool. Vibe Island monitors Claude Code, Codex CLI, ZCode, Gemini CLI, Antigravity CLI, Cursor Agent, Trae, OpenCode, MiMoCode, Factory Droid, Qoder, Qwen, Kimi Code, DeepSeek, Mistral Vibe, Copilot, CodeBuddy, WorkBuddy, Kiro, Hermes, Amp, Pi Agent, Oh My Pi, and Gajae Code.

Beyond a usage-only menu bar tracker

Most tools in this category do one thing: show you a usage number in the menu bar. That's useful, but it's a dashboard you glance at — not a system that acts. Vibe Island is a command center for your coding sessions. It shows live status, yes, but it also lets you approve permissions, answer questions, review plans, and jump straight to the work, all without leaving what you're doing. The difference is between watching a number tick and actually staying in control of every agent you've set loose.

Common questions

What is AI coding session monitoring?

It's a live view of every AI coding agent you have running — what each one is doing right now (working, waiting on you, or finished), surfaced in one place so you never lose track of a session or miss the moment one needs your input.

Does it track usage and cost too?

Yes. Alongside each session's status, Vibe Island shows your remaining subscription quota and window reset for each provider, plus token and cost breakdowns by day, month, and model. Usage and session monitoring share the same notch panel.

Which AI coding agents does it support?

Vibe Island monitors 25 AI coding agents, including Claude Code, Codex, Cursor, Gemini CLI, and OpenCode. New agents are added regularly, so the list keeps growing.

Does it read my code or conversations?

Vibe Island runs entirely on your Mac and reads only what it needs to show session status and usage. Your code and conversations never leave your machine — all monitoring stays local.

What Macs does it support?

Vibe Island runs on macOS 14 and later, on both Intel and Apple Silicon. It's native Swift and uses under 50 MB of RAM, so it stays out of the way while it watches your sessions.

Is it free?

Vibe Island comes with a 2-day free trial — every feature included, no card required. After that it's a one-time purchase, not a subscription.

Vibe Island app icon Vibe Island

Know what every AI coding agent is doing

macOS 14+ · Intel & Apple Silicon

Download Free Trial

2-day free trial · all features included · no card required