THE HUMAN AI OS
The system, one piece at a time.
I build my own operating system in public and release the pieces as I go. Every one is free, works in Claude, and needs no account, no install tooling and no API key.
03 live
The AI leadership team
Six executive seats for your own AI. A chief of staff, a CFO, a CPTO, a CISO, a CGO and a CMO, each with a real remit and a boundary it will not cross.
What is inside
- Six persona files, one per seat, plus a shared standard for how they answer
- An onboarding buddy that installs the whole team into a Claude Project for you
- A worked transcript, so you can see it run before you run it
ZIP, 6 files
Anti-AI proofreader
Strips the tells out of anything you write. It hunts the constructions that make good writing read like it was generated.
What is inside
- The skill file, one page
- A README with the two lines you need to start
- The list of tells it hunts, so you can argue with it
Helicopter view
A step-back partner for when a piece of work is drifting into something bigger than it needed to be.
What is inside
- The skill file, one page
- A README with the two lines you need to start
- The telltales it watches for, so it can fire before you ask
How to use any of them
Step 01
Download the zip and unzip it.
Step 02
Open the README. It is one page.
Step 03
Paste the .md into a new Claude chat. That is the whole install.
Coming
The decision ledger
Every decision you make, written down once, with the date you agreed to look at it again.
The memory system
How the system remembers what you told it six months ago, without you re-explaining yourself.
Scheduled runs
The work that happens while you are asleep, and how to trust it enough to stop checking.
Session rituals
Opening and closing a working session so nothing falls between two of them.