Deployment overview
Where the engine runs. Two options, the same engine.
Your factory is described in your repository. It runs wherever you start the engine.
Deployment options
| Option | What it is | Best for | Guide |
|---|---|---|---|
| Local | openfabriq installed with npm or Homebrew, started in a terminal | trying it, a solo project, your own computer | Local |
| Docker | one container with everything inside, configured by a .env file | a team, on an always-on server (VPS or cloud instance) | Docker |
Both run the same engine on the same openfabriq/ folder: start local, move to Docker when the team needs it. Already running Coolify? It deploys the same container, and a prompt can set it up for you.
How they compare
| Local | Docker | |
|---|---|---|
| Where | your computer, or any machine you manage | a server with Docker: Hetzner, OVHcloud, DigitalOcean, AWS EC2, Google Cloud, Azure… |
| When it works | while the machine is awake; a missed schedule is caught up once | all the time |
| Claude | your login (claude auth login) | an Anthropic API key, billed per token; or a claude setup-token token, for a server only you use |
| GitHub login | yours (gh auth login) | your team's GitHub App (openfabriq connect github), or a token of a bot account |
| Factories to run | every repository where you ran openfabriq apply | the GitHub repositories listed in OPENFABRIQ_REPOSITORIES, as owner/name |
| The studio | the address start prints, http://127.0.0.1:4310/?token=… | your domain and a password (OPENFABRIQ_STUDIO_URL, OPENFABRIQ_STUDIO_PASSWORD) or each person's GitHub account, or the printed address through an SSH tunnel |
Move from one to the other
Set up the engine on the new machine, then stop the old one. Never keep both: two engines on one repository each answer every event, which gives two pull requests for one issue.
Do you need the relay?
Only if the factory uses a service other than GitHub and schedules: Slack, Linear, Sentry or webhooks. The relay is a small service you deploy once, so that they can reach your factory while your machine stays unreachable from the Internet. See Relay.
After deployment
| To… | Read |
|---|---|
| follow what the factory does | Studio, or openfabriq status |
| understand a failed run | Troubleshooting |
| connect Slack, Linear or Sentry | Integrations |
| know what an agent can and cannot reach | Security model |