← Back

AiMessage

Your AI asks before it acts.

What it does

AiMessage sits between your AI and your computer. Before Claude Code can edit a file, run a command, or take any action — it texts you first. You reply YES or NO from your phone. That's it.

Install

npm install -g aimessage
aimessage setup

Setup asks for your phone number, installs a Claude Code hook, and sends you a test iMessage. Takes 30 seconds.

How it works

1. Claude Code tries to do something (edit a file, run a command)

2. The hook intercepts it and sends you an iMessage

3. You reply YES (always allow), ONCE, or NO

4. The action proceeds or gets blocked

5. A proof is recorded locally

Commands

aimessage setup

Install the hook and set your phone number.

aimessage status

Check if the hook is active.

aimessage proofs

Show recent proofs — every allow and deny.

aimessage uninstall

Remove the hook. Your proofs stay in ~/.aimessage/

Requirements

macOS with iMessage signed in. Node.js 18+. Claude Code.

How proofs work

Every decision (allow or deny) creates a proof — a SHA-256 hash that chains to the previous one. The chain is append-only. You can't delete or reorder entries. It's a local, cryptographic record of every action your AI took and every decision you made.

Proofs are stored in ~/.aimessage/proofs.json

Privacy

Everything runs locally on your Mac. No servers. No accounts. No data leaves your machine. Your phone number is stored in ~/.aimessage/config.json and is only used to send yourself iMessages.

Built on OCC (Origin Controlled Computing)

Proof Explorer →What is OCC →Whitepaper →Trust Model →Proof Format →Integration Guide →GitHub →