Viewerly

Blog · Guides · 16 Sep 2026 · 3 min read

How to use Viewerly with Claude Code

Point Viewerly at your projects, let Claude Code's hooks report status, and approve permissions without hunting for the right terminal.

Viewerly runs Claude Code sessions for you and keeps them all in one window. This is what that looks like in practice.

Add your projects

Open Viewerly and add a folder. If you keep everything under one parent directory, use the bulk option: Viewerly scans it for git repositories and adds them together, rather than making you pick them one at a time.

A project is just a folder. Sessions belong to projects, and the sidebar is an outline of both: projects at the top level, their sessions underneath.

Start a session

A session is one Claude Code process, running in its own terminal, in that project's folder.

Free runs one live session per project, in as many projects as you like. So a free Viewerly with twenty projects runs twenty agents, one each. Premium lifts the per-project limit, for when one project needs several agents working at once.

Status comes from hooks, not from the screen

Every session shows one of six states: idle, working, waiting for permission, waiting for input, done, or error.

Those come from Claude Code's own hooks, delivered over a local socket. Viewerly never guesses them by reading what scrolled past in the terminal. Terminal output is written for a person to read, it changes between versions, and parsing it produces a status that is confidently wrong. An agent that says it is waiting for you is waiting for you.

Install the hooks from Settings the first time. It writes them into your Claude Code configuration and tells you what it changed.

Approve permissions without switching

When Claude Code asks to run a tool, the request appears in Viewerly as a bar under the session header, with the exact command it wants to run. Allow or deny it there.

Every pending request across every project also lands in one queue on the Overview screen, and in the command palette. So the question is never "which of these terminals is blocked" — the answer is on one screen, and you can answer it from there.

The permission bar is deliberately not drawn inside the terminal. The terminal stays a terminal.

Work on the files while the agent does

Each project has a file browser, an editor and a git panel beside the terminal. You can read what the agent just changed, edit it yourself, and see the diff, without leaving for another window.

Reach it from somewhere else

Remote access is part of Premium, and off until you turn it on. When you do, pair a device and you can watch a session, approve permissions and send prompts from a browser.

What travels between your devices is encrypted end to end. The relay on viewerly.io stores ciphertext it has no key for; decryption happens on your own devices after pairing. That is a property of the design rather than a promise about our conduct: the server cannot read your sessions because it does not hold the keys.