Skip to content
Open source autonomous research desktop lab

Run your own autonomous research lab.

Install Maabarium, run a first workflow, then go deeper into blueprints, runtime paths, and release behaviour.
Local AI
Safe Git experiments
Persistent history
Real desktop console
See runs, blueprints, and outcomes at a glance
Shipping in public
Maabarium desktop console showing the console overview.
What you see instantly
Live score and progression
Why it matters
Decisions stay inspectable
Choose your path

Start where your question starts.

Get moving quickly if you are new, or jump straight to the product surface you need if you already know how you want to operate Maabarium.

Command strip

Keep the most useful commands visible.

This is the fast path for readers who already know what they are doing and just need the right command in front of them.

Install

Use the hosted installer for the latest public Apple Silicon desktop build on macOS 11+.

curl -fsSL https://downloads.maabarium.com/install.sh | sh
Open related docs
Run

Inspect the installed CLI entrypoint before running workflows from a terminal.

maabarium --help
Open related docs
Update

Rebuild from source after pulling the latest workspace changes.

git pull && cargo build --workspace --locked
Open related docs

This documentation is designed for two kinds of readers at the same time:

  • If you are getting started, follow the Start Here section in order.
  • If you are already comfortable with local AI tools, jump into the Guides and Reference sections.
  • If you are contributing content, head to Community.
Product areas

See the main operating surfaces before you drill into the docs.

These panels link the homepage directly to the parts of Maabarium that most users need to orient around: analytics, workflow configuration, and the persisted decision trail.

  • Run proposal, evaluation, and keep-or-reject loops in isolated Git-backed workspaces.
  • Keep decision trails visible through the desktop console, logs, persisted history, and runtime artefacts.
  • Work with blueprint-driven workflows instead of hand-wiring every experiment.
  • Stay local-first while still supporting multiple model and provider configurations.
Recommended blueprints

Start from bundled workflow templates instead of a blank page.

These summaries are pulled from the bundled blueprint templates, then paired with direct links to the source files so readers can move from explanation to exact configuration.

Open-source jump links

Use the docs site as the hub, then jump into the repo when you need source-level detail.

  1. What Maabarium is
  2. Install
  3. Run your first workflow
  4. Desktop console
  1. Workflows and blueprints
  2. Models and providers
  3. Updates and release channels
  4. App data and paths
  1. Contributing to docs
  2. Open the repository linked in the header.
  3. Use the pull request template and the preview build to validate changes.
Edit page

Last updated: