Free during early access — Mac app available now

Watch, steer, and approve every AI coding agent —
without sitting at the terminal.

Your agents don't stop when you step away from the terminal — one stuck on a question just sits idle until someone notices.

One calm place to follow every running agent — Claude Code, Codex, Gemini, and Qwen — on your own servers. Read the gist of each long session instead of the raw scrollback, jump in mid-run, and approve the big moves before they land — from your Mac or your phone.

Download for Mac — free during early access Request early access

Version 0.5.17 · Requires macOS 15 or newer · Sign in with an Expertly account — new here? Request access →

Signed and notarized. Sign in with your Expertly account and you're in — no credit card required. Don't have an account yet? Join the waitlist and we'll get you set up.

For developers and teams who run Claude Code, Codex, Gemini, and Qwen agents on their own servers — and want to keep an eye on them without being chained to the terminal.

How it works

Set up in minutes. Nothing new to install on your servers.

1

Nothing to run on your machines

There's no iVibecode server or agent to deploy. It connects to wherever your agents already run — your Mac, a cloud box, a server — over plain SSH. Nothing new is exposed to the internet.

2

Connect with the SSH access you already have

Open the app, point it at your machine, and connect using your existing SSH keys or your ~/.ssh/config. No new passwords to invent; it never edits your SSH config.

3

Watch your agents work

See live activity from every session — Claude Code, Codex, Gemini, or Qwen. Open any one to read the full log, send it a message, or approve a pending action.

4

Stay in control, off your laptop

Let agents run long jobs overnight or while you're out. You're notified the moment one needs you — and you can handle it from your phone in seconds. On the Mac you can even dictate your reply instead of typing it.

The reading view

Your terminal, finally readable.

Running several agents means several firehoses of reasoning you don't need to read. iVibecode's reading view distills each long session down to what actually happened and what it's asking you — so you can follow ten efforts at once instead of scrubbing ten terminals. It's the thing our own team reaches for first.

Raw terminal scrollback
User Can you make the pricing section clearer and check the release checklist? OpenAI Codex I will update the early-access copy so a new team can understand the offer quickly. (working through the landing page) Tool activity - Read static/index.html - Edited the pricing section - Checked the release checklist Waiting for you Approve publishing this demo copy to the staging page?
The same session, reading view
A real staged iVibecode reading-view screen showing synthetic Codex work, tool activity, and a waiting-for-you approval prompt.

See your whole fleet at a glance — every agent, every machine

Claude Code
feat/auth-refactor
Needs you
Tests passing — waiting to open the pull request.
OpenAI Codex
fix/rate-limiter
Working
Reproducing the burst-limit failure before patching.
Gemini
docs/api-reference
Done
Regenerated the endpoint docs — ready to review.
Qwen
test/flaky-suite
Working
Hunting the flaky integration test before re-running the suite.
One window onto your whole fleet

Claude Code, Codex, Gemini, and Qwen —
in one place you can actually read.

Run Claude Code in one repo, Codex in another, Gemini and Qwen somewhere else, and you're checking four different things. iVibecode shows all of them together, labels each one correctly, and lets you jump into any session and answer it.

C

Claude Code

The most-polished path — every long session rendered as a readable gist with its own per-agent parser.

OpenAI Codex

Read alongside the rest, labeled "OpenAI Codex" — not mislabeled as Claude. Its own dedicated parser.

G

Gemini

Same readable view, correctly labeled "Gemini," so your whole fleet lives in one window.

Q

Qwen

Qwen sessions are recognized and labeled. Qwen Code is a Gemini fork, so its sessions read through the same parser as Gemini.

Why this beats a free phone remote

Most free phone remotes are built around a single vendor's agent — usually Claude. If you also run Codex, Gemini, or Qwen, a single-vendor remote leaves those sessions out. iVibecode's whole point is the opposite: see your fleet, not just one vendor's agent — every running agent in one readable view.

Security

Your servers. Your keys. Your repositories stay on your servers.

A team won't tunnel a tool into its own infrastructure without a clear security story. Here's the honest one.

Connects over the SSH access you already trust

iVibecode uses plain SSH to reach your machines — the same secure connection you already use for remote work. It works with your existing SSH keys and your ~/.ssh/config, and it never edits your SSH config.

Nothing new exposed to the internet

There's no iVibecode server or agent to install on your machines, so there's no new open port or service to harden.

Your repositories stay on your servers

The agents — and your repositories — run where they already run. iVibecode connects over SSH and never copies your repositories. One honest caveat: the AI status summaries send the last screenful of session text — which can include code that was on screen — to Expertly's AI service to write the gist.

Credentials handling

SSH keys are kept in a private, permission-locked store on your device, and the password-setup path installs your device key and does not keep the password.

Pricing

Free during early access.

Early users lock in founder pricing
Free while we're in early access

Download the Mac app and use it free during early access. Team plans are coming — and the teams that start now keep founder pricing when they launch.

Why trust it

Built by people drowning in their own agents.

We were drowning in terminal windows running our own agents, so we built the view we wanted: one calm, readable place to see every session, understand what each one is doing, and answer the ones waiting on us — from anywhere.

— The iVibecode team
An Expertly product
Works with Claude Code OpenAI Codex Gemini Qwen
FAQ

Questions, answered.

Which agents does it work with?

Claude Code, Codex, Gemini, and Qwen, running in your terminal sessions — each gets its own readable view. (Claude Code, Codex, and Gemini each have a dedicated parser; Qwen Code is a Gemini fork, so Qwen sessions ride Gemini's. The Claude Code path is the most-polished.)

How long does setup take?

Minutes. There's nothing to install on your servers — you connect with the SSH access you already have. If a machine is already in your ~/.ssh/config, the details auto-fill.

Where does my data go?

Your agents and your repositories stay on your machines — iVibecode connects over SSH and never copies your repositories. The AI status summaries send only the last screenful of session text to Expertly's AI service to write the gist. (Credential specifics: see the security section.)

What do I need to run it?

A Mac to run the app (iPhone and iPad are in early access too), and at least one server you can reach over SSH with your agents running on it. That's it — no new server software.

What does it run on?

The Mac app is available now (signed and notarized), and it keeps itself up to date — new versions download and install automatically. iPhone and iPad are in early access — join the waitlist. A browser version that needs no Mac is in the works.

Is it free?

Free during early access. Team plans are coming; early users lock in founder pricing.

Your agents are running. See them.

Download the Mac app and start watching your fleet in minutes — free during early access. Need an Expertly account to sign in? Request early access below and we'll set you up. iPhone and iPad are in early access too — the same request gets you a TestFlight invite.

Download for Mac — free during early access
Thanks — you're on the list. We'll email you with early-access details and get your Expertly account set up.

No spam, ever. Unsubscribe at any time.