Personal operating system
AI OS
The AI system I run my own work and life through.
I built it to run my own work. There is a vault of plain markdown files holding who I am, what I am working toward, and the decisions I have made and why. On top of that sits a set of skills that read the vault and actually do things with it.
Two layers
The vault is just files I own. Identity, goals, the decisions I have made, how I work, even notes on how I write. Nothing locked inside someone else's app.
On top of it sit the skills, about twenty of them now. They run the morning brief, draft content, do the weekly review, and audit the system itself.
Safe-fail by default
Nothing reaches the outside world unless I have handed it a specific key for that. Without one, a skill can draft an email but not send it, or stage a change but not apply it.
The destructive actions stay off until I switch them on myself. That is the only reason I am happy to let the thing run while I am not watching.
Reversible by design
Every action writes a line to an audit log and commits to git. So the whole thing is undoable.
If a routine does something I did not want, I can see exactly what it did and roll back to where things were before.
Stack
It runs on my own machine against my own files, so there is nothing for you to log into. This page is the closest thing to a demo.