browser

What This Skill Does

The Browser skill for OpenClaw provides a robust interface for automating web browser interactions using natural language. Built on top of the Stagehand CLI, this tool allows you to perform complex web navigation, form filling, data extraction, and UI interaction without needing to write Selenium or Playwright scripts manually. It seamlessly switches between a local Chrome instance for rapid development and a remote Browserbase environment for production-grade, stealth-capable scraping that handles proxy management and CAPTCHAs automatically. By abstracting away the underlying browser drivers, this skill empowers you to interact with the web as if you were guiding a human assistant, simply by stating your objective in plain English.

Installation

To install this skill, use the ClawHub CLI: clawhub install openclaw/skills/skills/peytoncasper/browser-pc. Before using, ensure your environment is prepared by checking the setup.json file. If setupComplete is set to false, run npm install followed by npm link in the skill directory. For remote execution, configure your .env file with BROWSERBASE_API_KEY and BROWSERBASE_PROJECT_ID. If these keys are absent, the skill will default to a local Chrome browser. Always ensure Chrome is installed on your local machine if you plan to use local mode.

Use Cases

This skill is ideal for automating repetitive web tasks. Use it to extract structured data from e-commerce sites, monitor price changes on competitor websites, fill out registration forms, or perform automated testing of web applications. It is particularly valuable for research tasks where you need to scrape multiple pages, take screenshots for documentation, or verify the layout of a web application across different sessions.

Example Prompts

1. "Open https://www.google.com, search for 'latest AI trends', and take a screenshot of the results page."
2. "Go to the contact page of this company and extract the email address and phone number into a JSON object."
3. "Navigate to the login page and fill in the username 'admin' and password 'secret', then click the submit button."

Tips & Limitations

Always start by navigating to the target URL before attempting interactions. If an action fails, use the browser observe command to see how the AI perceives the page elements; this is often better than retrying the same command. Note that local execution lacks stealth features, so it may trigger anti-bot measures on sensitive sites; switch to Browserbase mode for production scraping. Remember to execute browser close to free up resources once your task is complete to ensure your system remains performant.