geass

Absolute command
over servers you already own.

One cloud to manage them all.

Your own cloud on your own servers, from VPS to bare metal. Ship apps and databases, scale across machines and connect your other clouds, all in one place, with no lock-in.

Install Geass Read the docs
The Geass dashboard: a project's canvas of apps, a function, databases and a bucket with their status, the environments across the top with a Deploy changes button, and the selected database's details on the right.
ONE CLOUD

One control plane over everything you run

Your servers and the services you already use, managed in one place, by you or by your agents.

DRIVEN BY
  • Dashboard
  • API
  • CLI
  • MCP

Servers at more than one provider in a single cluster is experimental.

Docs: Add servers · Cloud connectors

GET STARTED

Sign the contract in one line

Three steps from a fresh server to a running cloud. It all runs on your own hardware, so your data stays yours.

  1. 01

    Install

    Run one command on a fresh Linux server. Geass sets up the cluster.

  2. 02

    Connect

    Join more servers, then plug in your code and your cloud accounts.

  3. 03

    Ship

    Push code or pick a template. Geass runs it, backs it up and keeps it running.

linuxsudo $curl -sfL https://github.com/degoke/geass/releases/latest/download/install.sh | sudo sh

2 GB RAM · amd64 or arm64 · root access · read the install guide

WHAT YOU GET

A managed cloud, on servers you own

Everything you would expect from a cloud platform, running on your own VPS and bare-metal machines.

SHIP

Push and it ships

Deploy from GitHub or any container image. Every pull request gets its own preview.

  • Functions
  • Templates
  • Your own domain
  • Object storage
Docs Apps GitHub and builds Pull-request environments
DATA

Databases that look after themselves

Postgres, MySQL, Redis and SQLite on your own servers, with backups and restore built in.

  • Point-in-time restore
  • Shared servers
  • 7-day undelete
  • SQL console
Docs Databases Backups and restore Shared servers
SCALE

From one server to a fleet

Join another VPS or bare-metal machine and your work spreads across it.

  • Regions and tags
  • Autoscaling
  • HA Postgres
  • Replicated volumes
Docs Add servers Scaling and placement High availability
REVIEW

See the diff, then deploy

Edits wait as drafts. Review exactly what changes, then ship it all at once.

  • Instant live settings
  • Capacity checks
  • Deploy history
Docs Projects and environments Core concepts
OBSERVE

See what it is doing

Live logs, metrics, a shell and a SQL console, right in the browser.

  • Live logs
  • CPU, memory, network
  • Terminal
  • Signed webhooks
Docs Observability Usage and capacity
OWN

Yours to keep

Self-hosted on your own hardware, with no telemetry and no lock-in.

  • Private by default
  • Admin and viewer roles
  • Source-available
  • Standard Kubernetes
Docs Security Users and tokens Network isolation
AGENTS AND AUTOMATION

Your agents can run it too

A command line, an MCP server and a versioned API. Give an agent a token and it can ship, debug and fix your cloud, inside the limits you set.

  • SHIP

    Deploy a service and wait until it is ready, or roll back a bad release.

  • DEBUG

    Read status, metrics and deployment history, and see what the last deploy left behind.

  • FIX

    Change variables and secrets, scale a service, or let it sleep when idle.

  • SCHEDULE

    Give a function a schedule, and call it on demand to check it.

  • PLAN

    See nodes, regions and compliance tags, and how to reach a database.

  • GROW

    Create a service from an image or a GitHub repository, as a draft you review.

$ geass status thumbnails
NAME        STATE   READY  BUILD  URL  MESSAGE
thumbnails  failed  0/1    -      -    CrashLoopBackOff

$ geass env set thumbnails MAX_IMAGE_MB=8
$ geass deploy thumbnails --wait
thumbnails: Deploying thumbnails
thumbnails: ready at https://thumbs.example.com

youWhy does thumbnails keep crashing?

agentapp_status(app: "thumbnails")

agentapp_metrics(name: "thumbnails", window: "1h")

agentset_variable(app: "thumbnails", key: "MAX_IMAGE_MB", value: "8")pending

agentIts memory sits at the limit and it restarts whenever someone uploads a large image. I capped uploads at 8 MB as a pending change. Want me to deploy it?

youYes, go ahead.

agentdeploy_app(app: "thumbnails", wait: true)ready

$ export GEASS_TOKEN=gsp_…
$ curl -H "Authorization: Bearer $GEASS_TOKEN" \
"https://geass.example.com/api/v1/apps?project=storefront"

# Every route is described at /api/v1/openapi.json,
# and signed webhooks tell you when a deploy changes.

One token, the same limits, whichever way an agent connects. CLI docs · MCP docs

Scoped by project

A project token reaches only that project, so an agent touches only what you give it.

Read-only when you want

A read-only token can inspect everything it is allowed to see, logs included, and change nothing.

Drafts, then your approval

Changes wait as pending until you deploy, and a delete needs the resource's name typed back.

Terms of the contract

Elastic License 2.0
YOU KEEP
  • Your servers
  • Your data and backups
  • Your domain
  • Your bill
GEASS HANDLES
  • Builds and deploys
  • Routing and domains
  • Backups and restores
  • Restarts and placement

Source-available under the Elastic License 2.0. Read the license.

Sign: install Geass