OpenClaw

Full Guide

OpenClaw is an open-source, self-hosted personal-assistant gateway that connects model providers, messaging channels, tools, and an operator-controlled workspace.

What it is

The gateway can run on a local machine, NAS, server, or container host. It is not itself a model. Operators connect it to a provider such as Ollama for local inference or to a hosted model API for cloud inference. A self-hosted gateway therefore does not guarantee that every prompt or tool result stays local.

Deployment and security

The official Docker path supports Docker Engine plus Compose v2 and publishes versioned images through the project’s GitHub Container Registry. The setup creates persistent configuration, workspace, and authentication-secret directories, performs onboarding, and starts the gateway.

OpenClaw’s security documentation assumes one trusted operator boundary per gateway. Its recommended controls include strong gateway authentication, narrow channel allowlists, limited tool and filesystem authority, sandboxing where appropriate, and a security audit after configuration changes or before network exposure.

Local models

OpenClaw supports Ollama in local-only, cloud-only, and hybrid modes. A private Ollama endpoint can keep model inference on the operator’s own hardware, but model quality, context, tool reliability, and prompt-injection resistance still depend on the selected model and hardware.

Kingy coverage

Read Kingy’s evidence-led OpenClaw-on-UGREEN-NAS setup and security guide.

Official sources

• OpenClaw documentation
• Docker installation
• Gateway security
• Ollama provider guide
• Official source repository