Network devices

Catalyst runs in the cloud, but a lot of what you might want it to use runs somewhere private — a model on your laptop, a service on a workstation, an API inside your company network. The cloud can’t reach those by default. Network devices bridge that gap by joining your machine to a secure private network Catalyst can see.

It uses Tailscale, which creates a private, encrypted network between your devices — so “your laptop” becomes an address Catalyst can reach, without opening anything to the public internet.

Connecting a device

1. Open Network from the sidebar and click Add Device.

2. Paste a Tailscale invite URL for the device you want to connect.

   

3. Accept the invite on the machine. It joins the private network and becomes reachable from Catalyst.

What it’s for

Once a machine is connected, you can point Catalyst at services running on it:

• A local model via Ollama or any OpenAI-compatible server.
• A private MCP server or tool (see Integrations).
• Any internal HTTP service a workflow HTTP node needs to call.

Note

This is the piece that makes “self-hosted” practical: keep the model or service on hardware you control, and still drive it from the same Catalyst chat and workflows as everything else.