A stateless CLI client that allows AI agents to control real, authenticated remote browsers through simple execution commands.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install claw-relay-browser
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install claw-relay-browser using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
Claw Relay is a specialized interface designed for AI agents that interact with the web through direct command execution. Unlike traditional browser automation that relies on headless environments, this skill allows Openclaw Skills to leverage a user's real browser session. This means the agent can access sites where the user is already logged in, bypassing complex authentication hurdles by using existing cookies and active sessions.
The tool is built to be lightweight and stateless. Instead of maintaining a heavy, persistent connection, the agent makes individual calls to a Node.js script to perform specific tasks. This architecture makes it highly compatible with Openclaw Skills that prioritize reliability and simplicity in remote browser orchestration.
To integrate this with your Openclaw Skills, first ensure you have Node.js and the necessary dependencies installed in the script directory.
npm install ws
Configure your environment variables to authenticate with the relay server:
export CLAW_RELAY_URL="wss://relay.clawrelay.dev/"
export CLAW_RELAY_TOKEN="your-token"
export CLAW_RELAY_AGENT="your-agent-id"
You can then execute actions directly:
node relay-client.cjs navigate https://github.com
The skill returns structured JSON data for every action performed. Below is the typical schema for snapshots and interactions:
| Attribute | Description |
|---|---|
ok |
A boolean indicating the success of the CLI command. |
snapshot |
A textual representation of the page's accessibility tree with element references (e.g., e1, e2). |
path |
The file path where a screenshot was saved, if applicable. |
error |
Detailed error messaging if a scope violation or connection issue occurs. |
result |
The return value from JavaScript evaluation actions. |
Loading
A professional technical research skill that orchestrates multi-source evidence for framework selection and technology comparison.

An AI-driven sprite generation and processing workflow designed to turn text prompts into engine-ready pixel art assets for Godot within the Openclaw Skills ecosystem.

An advanced AI research agent that builds structured B2B prospect profiles, identifies key decision-makers, and generates personalized engagement recommendations.

A specialized tool that transforms vague reporting clues into structured, data-driven news pitches using knowledge graph context.

An automated production pipeline that transforms a single topic or idea into a fully edited 60-90 second vertical video delivered via Telegram.

An AI-powered creative tool for generating retro-aesthetic, synthwave, and vaporwave images using simple text prompts.








































