OpenClaw Glossary
Understand OpenClaw basics, core agent terms, and setup concepts before you jump into hands-on guides.
Core concepts to learn first
Use the glossary to understand OpenClaw, gateway tokens, browser relay, channels, and skills before you change configs or deploy to a VPS.
A-Z glossary
Browser Relay
Browser relay is the component or workflow that lets OpenClaw safely bridge browser-driven actions or sessions back to the gateway and your configured environment.
Channel
A channel is the user-facing communication surface OpenClaw connects to, such as Telegram, WhatsApp, or Feishu.
Fallback
Fallback is the backup handling strategy when the primary model or route fails, times out, or violates policy.
Gateway
The OpenClaw gateway is the service layer that receives channel events and routes them to configured agents.
Gateway Token
A gateway token is the secret used to protect access to the OpenClaw gateway or related control endpoints. It should be stored like any other high-sensitivity credential.
OpenClaw
OpenClaw is an open-source AI agent runtime and tooling ecosystem that combines channels, gateway routing, models, skills, and operational controls.
Routing
Routing defines which model or agent should handle a request based on channel, intent, or policy constraints.
Skill
A skill is a reusable capability, workflow, or extension that OpenClaw can install, call, or expose through its environment.
Tunnel
A tunnel exposes your local or private service to the public internet over a secure endpoint for testing callbacks.
Webhook
A webhook is an HTTPS endpoint used by external platforms to push real-time events to your OpenClaw instance.