Getting Started

There are three ways to use Octo, and they all execute the same flow YAML. Pick the one that matches how you want to work today — nothing you build is locked to any of them.

The three ways to run Octo

The octo CLI. The runtime is a single Go binary. Point it at a config file and it starts your connectors and flows; add --watch for hot reload while you edit, or use octo invoke to call a single flow with JSON in and JSON out. This is the best fit for local development, testing, and CI.

The standalone visual editor. A published Docker image bundles the editor and the octo runtime. You mount a directory of flow YAML, design flows on a visual canvas in the browser, and run them with one click — no build, no database, no setup beyond docker run.

The full platform on Kubernetes. A control plane (orchestrator plus Postgres) and a multi-user editor that store integration definitions and deploy each one as its own Kubernetes workload, with environment binding, secrets, scaling, versioned deployments, and centralized logs. This is how you operate Octo in production as a team.

Where to begin

If you are new to Octo, install the CLI and run your first flow — it is the fastest way to understand the model. If you would rather not install anything, start with the editor quickstart instead.