Skip to content

Install & Access

Supen is easiest to learn when you do one thing first: get into a working space, then start a thread from the platform or the CLI. This page covers the fastest ways to do that.

If your team uses the hosted deployment, open the Platform link in the docs navbar or go directly to https://app.supen.ai.

If you are running Supen yourself, start with the CLI:

Terminal window
supen bootstrap
supen start
supen status

supen bootstrap handles first-time setup. supen start brings the local service online. supen status confirms that the service and attached tools are reachable.

If you are working from this repo, start the app and daemon together:

Terminal window
pnpm dev

Local development entrypoints:

  • Platform: http://localhost:2655
  • Daemon: http://localhost:2756

Once Supen is running, open the platform and:

  • choose a connected computer if one already exists
  • connect your first computer if the workspace is empty
  • pick the computer you want to work in before starting threads

The platform connection flow gives you a one-time command to run on the machine you want Supen to manage.

If you prefer terminal-first work, the CLI can connect to the same runtime:

Terminal window
supen status
supen chat "hello" --agent web-agent

Useful first commands:

  • supen status checks the service, channels, and detected coding CLIs
  • supen chat sends a task to an agent and streams the reply
  • supen computer list shows computers available through the gateway
  • supen computer launch <name> launches a provisioned computer from the CLI when your gateway supports provisioning

You are ready to continue once you can do all three of these:

  • open the platform and see a computer
  • run supen status without a connection error
  • start or resume a thread from either the platform or the CLI