Skip to main content
Creating a brain is one command. Two things about that command cannot be changed afterwards, so onboarding is mostly about deciding them on purpose — and vitruvio refuses to decide either one for you.
1

Decide: governed or ungoverned

Whether a trust root travels with the brain and signatures gate what consumers accept. This lives in the genesis and cannot be added later. Governed or ungoverned.
2

Decide: standalone or in a project

One brain, or several under one configuration with one repository each. Create a brain explains the choice and gives the four recipes.
3

Create it, then check it

One command, then brain verify and auth status, read separately: integrity and authenticity are different questions.
4

Register evidence and search

source register, index build, search. First evidence and search.

​
Or let your agent drive

Install the skills and ask your agent to set up a brain. The vitruvio-onboarding skill runs this same flow, and it is written to stop and ask you both decisions before it creates anything — the CLI’s defaults are never taken as your answer.

​
What you have at the end

A directory that is an OCI image layout, a vitruvio.toml beside it (or the project’s, one level up), one snapshot with zero blocks, and an actor recorded for every write to come. vitruvio brain state --json shows all of it, and is the command to run first in any new session from now on.