What Libro gives Claude Code.
Libro is the operating layer we run Cerebro on, packaged so you can run it too. You bring your Claude. Libro brings the structure around it.
A Brain that remembers
Sessions, decisions and an awareness layer, as Markdown files in a folder you own.
Session rituals
A grounded brief opens each session. A structured retro closes it.
A doctor
One command checks your skills, your scaffold and your dependencies.
Provenance lint
Every memory file carries an 11-key provenance header, and the lint enforces it.
Skills you can read
Each skill is a SKILL.md file. Read it, change it, or manage it with skill-manage.
Three starter skills
daily-brief, decision and memory-recall, in the libro-starter profile.
Four things you keep.
Your memory.
Decisions, sessions and notes live on your disk, in a folder you choose. No vendor holds them.
Your rituals.
How a session opens and closes is a file you can edit, not a setting you cannot see.
Your skills.
Every skill is readable Markdown and scripts. Nothing runs that you cannot inspect.
Your Claude.
Libro ships no model. Bring your own Claude Code or claude.ai subscription.
Free. And it stays on your machine.
Libro is free. Not freemium. Not a trial. There are no Libro servers, so your data never reaches us. Files you can read, paths you can grep.
Uninstall anytime. It removes only what Libro installed, backs that up first, and leaves your own files where they are.
Need the governed, on-premises version, with a policy gate and an evidence ledger? That is what Cerebro delivers.
Trustworthy by default.
Reversibility
An install over an existing Brain keeps a backup, and one command restores it. Records are superseded, never overwritten.
Observability
Each install writes an activity log, and every session leaves a note you can read.
Reproducibility
Every memory file carries an 11-key provenance header, and a lint enforces it.
Install Libro. Keep your stuff.
Needs macOS or Linux, Bash 4 and Python 3.10 or newer, and your own Claude subscription.
Open source under Apache 2.0. Read the script before you run it. That's the point.
Build notes
Short notes on what we build, what breaks and how we fix it. No spam.
Straight answers.
What is Libro?
Libro is a free, open-source operating layer for Claude Code. It gives Claude Code a Brain that remembers, session rituals and skills you can read. Everything lives in local files you own.
Is Libro free?
Yes. Apache 2.0, not freemium and not a trial. You bring your own Claude subscription. Libro is the structure you run it inside.
How is Libro different from an AI model?
A model answers questions. Libro is the structure around it: memory you can read, rituals that keep each session grounded, and skills you can inspect. It ships no model of its own.
Does Libro send my data anywhere?
No. There are no Libro servers. Your memory is local files you can read and grep.
Does Libro route between models?
No. Version 0.3.0 ships no model routing. The three dispatch skills in the full profile are retired and refuse live calls unless you opt in.
What do I need to run Libro?
macOS or Linux, Bash 4 and Python 3.10 or newer, and your own Claude Code or claude.ai subscription. Windows through WSL2 is untested.
Who builds Libro?
Cerebro Cyber Solutions, a service-disabled veteran-owned small business. We run our own company on the same conventions.
Where Libro fits.
Governed AI: what it means
The governed, on-premises version Cerebro delivers, and how to check our work yourself.
Cortex live demo
Watch a policy gate allow, hold or block an AI agent's spending, with receipts and ledger hashes.
What is a Claude Code framework?
How memory, skills and routing turn bare Claude Code into an operator-owned system, and which of those Libro ships.
AI that remembers you
Why most AI forgets between sessions, and how local memory keeps your context in files you own.
Own your AI, don't rent it
What operator-owned AI means in plain English, and why owning your memory and your rules beats renting a black box.