Open source / Company Brain OS

Company Brain OS

The open source company brain for AI agents. Markdown, git and Claude Code: no app to host, no database to run.

Free, MIT licensed, release v1.1.0. "I run my company on this brain. This is the open version." (Marco Galluccio)

What you get

A memory that survives anyone leaving

  • One loop: sync, work, debrief

    Every operator and every agent reads the same state before starting, and writes it back before leaving. Nobody keeps the context in their head.

  • Project memory, readable cold

    A flat index in memory/MEMORY.md plus one file per project, reference or lesson. A new operator or agent gets the current state without a handover meeting.

  • Daily logs per operator

    One file per person per day, so parallel sessions never collide. The history of what happened, searchable forever.

  • Git as the sync layer

    Plain markdown in a private repo you own. Direct push by default, branch protection and review gates when you need them.

  • A CLAUDE.md per folder

    Every folder tells the agent how that area works, so nothing gets read out of context.

  • 14 skills, all optional

    Guided Claude Code flows for the loop and the things you repeat: briefing, debrief, handover, meeting debrief, checkups, sparring.

How to start

Up and running in one session

  1. Create the repo

    From the template

    On GitHub, use this template to create your own private repository, then clone it to your machine.

    git clone https://github.com/<you>/<your-brain>.git
    cd <your-brain>
  2. Link the skills

    One-time setup

    Symlink the skills folder into .claude so Claude Code picks them up automatically.

    mkdir -p .claude && ln -s ../skills .claude/skills
  3. Boot it

    Inside Claude Code

    Start Claude Code in the repo and run the setup skill: it walks you through the rest.

    claude
    /setup-company-brain

Keep the repo private: your memory, logs and client notes live there. Needs git 2.31+ and bash; gh and Python only for the optional layers. Working alone, not a team? There is a solo version without the multi-operator scaffolding: Big Brain OS.

Want this for your team?

I teach teams to work with AI agents, in-company or in class in Rome.