Geass documentation

Geass turns servers you already own, from a single VPS to a rack of bare metal, into a managed cloud. You ship apps and databases from one dashboard, a command line or an AI agent, scale across machines, back up your data, and connect the clouds you already use, while everything runs on hardware you control.

These docs describe what Geass does today: what each feature is for, how to use it, what it does under the hood, and what to check when something goes wrong.

Start here

What is Geass What it is, who it is for, and when it is the right tool.

Install Geass One command on a fresh server. Requirements, what the installer does, how to confirm it worked.

Deploy your first app From a fresh install to a running, reachable app.

Core concepts Projects, environments, resources, pending changes, nodes and connections.

Connect your clouds

Geass uses accounts you already have. Each one is a connector in Settings → Connectors. See Cloud connectors for the permissions each one needs.

AWS S3 buckets and database backups in your own AWS account.

GitHub Deploy on every push, with a preview environment for each pull request.

Cloudflare DNS for the dashboard and an optional Cloudflare Tunnel.

PlanetScale Managed MySQL and Postgres databases in your PlanetScale organization (experimental).

Build and deploy

Apps Run code from a Git repository or a container image.

GitHub and builds Connect repositories and deploy on every push.

Pull-request environments A throwaway copy of your project for every pull request.

Domains and TLS Put your apps and the dashboard on your own domain.

Data and storage

Databases Postgres, MySQL, Redis and SQLite, with the connect string wired into your apps.

Backups and restore Back up to your own AWS account and restore to a new database.

Object storage Buckets for your apps, in the cluster or in AWS.

Storage and volumes Local storage, the Longhorn add-on, and resizing volumes.

Run and scale

Scaling and placement Replicas, regions, and how workloads land on nodes.

Sleep when idle Stop a service while nobody uses it and start it on the next request.

Compliance node tags Keep a project's work on the servers you choose, and other projects off them.

Network isolation Projects are private by default. Open exactly what you need.

Logs, metrics and terminal See what your services are doing and get a shell in them.

Troubleshooting Symptoms, causes and fixes, in the dashboard and with kubectl.

Automate

Command line Install geass, sign in with a code you approve in the dashboard, then deploy, watch and scale from a terminal or CI.

MCP server for AI agents Let an AI agent inspect and run services as an API token, limited by that token's role.

Users and API tokens Dashboard roles, and tokens for scripts, CI, the command line and agents.

Reference

REST API The versioned /api/v1 and the dashboard routes, with API tokens.

CLI commands Every geass command, flag, environment variable and exit code.

MCP tools Every tool the geass MCP server offers an agent.

Resources Every custom resource, field by field.

Configuration Installer variables, operator flags and platform settings.

Glossary The words Geass uses, in one place.