Claude Remote
One inbox for every Claude Code session you run. Permission prompts from all your machines land on your phone - approve, deny, or steer any AI coding agent from anywhere.
Built for the way we actually run the work.
One inbox, every session
Every permission prompt from every Claude Code session on every machine lands in a single live inbox, labeled by project and device.
Approve from your phone
An instant push tells you what a session wants to run. Approve, deny, or deny with an instruction the agent reads and acts on.
Steer work mid-flight
Queue a follow-up instruction to any live session and it is delivered the moment the current turn finishes - the agent keeps working instead of stopping.
Teach it with rules
Tap Always once and matching requests are approved server-side in about a second - no push, no card, no interruption.
Fail-open by design
If the server is unreachable or you do not respond in time, the prompt simply falls back to the terminal. Your work never blocks on the tool.
All your machines
A one-command install covers every project on a machine. Each device gets its own token you can revoke independently from the app.
From install to first approval in minutes.
Install the hook once per machine
A single command registers Claude Remote user-wide. Every Claude Code session on that machine reports in - no per-project setup.
Prompts come to you
When a session needs permission it holds while a push hits your phone with the project, the machine, and the exact command.
Decide from anywhere
Approve or deny in one tap, or reply with an instruction. The session continues in about a second - or falls back to the terminal if you are busy.
Let rules take over
Approve once with Always and Claude Remote handles that request type for you from then on, with a full audit trail in History.
Want to use Claude Remote for your team?
We license the cazyweb platform to selected partners and run it for our own clients. Tell us what you need and we will set you up.