A visual, human-like web browser skill that gives AI agents the ability to render JavaScript and capture screenshots.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install claw-browser
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install claw-browser using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Claw Browser?
The Claw Browser skill provides agents with visual capabilities by utilizing a real Chromium browser powered by Playwright. Unlike traditional HTTP clients, this addition to Openclaw Skills handles complex Single Page Applications (SPAs), dynamic content, and JavaScript-heavy websites, ensuring the agent interacts with the web exactly like a human user would. It essentially gives the agent eyes to navigate the modern web effectively.
Claw Browser Use Cases
- Analyzing websites that require full JavaScript rendering like React, Vue, or Angular apps.
- Capturing visual evidence through high-quality screenshots of web pages for verification.
- Interacting with dynamic content that fails to load via standard static HTTP requests.
- Demonstrating browser actions to users in visible mode during live sessions or debugging.
- Accessing web data that is protected or hidden behind client-side rendering logic.
How Claw Browser Works
- The agent initializes the browser environment using the provided Python scripts within the Openclaw Skills ecosystem.
- A URL and specific action, such as read or screenshot, are passed as arguments to the execution tool.
- Playwright launches a Chromium instance, either in a hidden headless state or in visible mode for user observation.
- The browser fully renders the target page, executes all necessary scripts, and performs the requested action.
- Extracted text or generated image files are returned to the agent for further processing or reporting.
Claw Browser Setup
To get started, ensure you have Python installed, then run the following commands to prepare your Openclaw Skills environment:
# 1. Create virtual environment
python3 -m venv venv
# 2. Activate it
source venv/bin/activate
# 3. Install dependencies
pip install -r requirements.txt
# 4. Download Chromium binaries (CRITICAL STEP)
playwright install chromium
Claw Browser Data Schema & Taxonomy
The skill manages data through direct command-line arguments and specific file outputs to maintain clarity within Openclaw Skills:
| Component | Details |
|---|---|
| Input Arguments | URL (Required), Action (read/screenshot), Visible (boolean) |
| Read Output | Markdown-friendly text content extracted from the rendered DOM |
| Screenshot Output | A local image file saved as evidence.png |
| Execution Environment | Local virtual environment (venv) with Playwright Chromium binaries |
Claw Browser Advanced Features
- Visible mode execution allows developers to watch the agent's browsing process in real-time for demos.
- Cross-platform support ensuring compatibility across Darwin, Linux, and Win32 environments.
- Headless mode for efficient background processing without interrupting the user's workspace.
- X-Server display configuration support for running visible browsers within WSL or Linux environments.
- Integration-ready architecture designed to work seamlessly with other Openclaw Skills for complex multi-step automations.
SKILL.md
Loading
Related Openclaw Skills
Storage Cleanup
A zero-dependency, one-command cleanup utility that reclaims gigabytes of disk space by purging system caches, logs, and build artifacts.
ClawGuard Task & Gateway Watchdog
A lightweight system-level watchdog designed to monitor sub-agent processes and automate configuration recovery for the OpenClaw gateway.
ddgs-search
A zero-config, multi-engine web and academic search tool that provides clean JSON results without requiring API keys or subscriptions.
ddg-search
A versatile command-line tool for scraping DuckDuckGo search results into multiple structured data formats including JSON, CSV, and Markdown.
Wiz Light Control
A local network integration that enables AI agents to control physical Wiz smart bulbs through Python-based CLI commands.
cancorteaw-app
A secure and controlled runner for scaffolding, developing, and previewing Expo/React Native applications directly within your environment.
Featured*
