use-browser

name	use-browser
description	Enable browser use with Playwright Python. Use this skill when the user asks to access or interact with a web page, scrape information from a website, automate web tasks, or perform any action that requires a real browser environment. This skill allows you to navigate websites, fill out forms, click buttons, and extract data as if you were using a browser yourself.

This skill enables browser use with Playwright, allowing you to access and interact with web pages, scrape information, automate tasks, and perform any action that requires a real browser environment.

Installation

pip install playwright --break-system-packages -q && playwright install chromium 2>&1 | tail -5

Scripts

Ready-to-use scripts live in claude/skills/use-browser/scripts/. Run them directly with python3 scripts/<name> or copy/adapt as needed.

All scripts save output under /tmp/web/<site-name>/.

Script	Purpose	Usage
screenshot	Full-page screenshot	screenshot <url> [output_path]
scrape-text	Extract visible text	scrape-text <url>
scrape-html	Extract full HTML	scrape-html <url>
scrape-links	List all links (href + text)	scrape-links <url>
wait-and-scrape	Wait for selector, extract text	wait-and-scrape <url> <selector>
multi-page	Paginate and collect items	multi-page <url> <item_sel> <next_sel> [max_pages]
auth-session	Login and reuse saved session	auth-session login <url> <user> <pass> / auth-session use <url>

Patterns

Headless vs headed

browser = await p.chromium.launch(headless=True)   # default, no window
browser = await p.chromium.launch(headless=False)  # visible window for debugging

Intercept network requests

await page.route("**/*.{png,jpg,gif}", lambda r: r.abort())  # block images
await page.route("**/api/**", lambda r: print(r.url) or r.continue_())

Execute JavaScript

result = await page.evaluate("() => window.__SOME_GLOBAL__")
await page.evaluate("window.scrollTo(0, document.body.scrollHeight)")

Handle dialogs

page.on("dialog", lambda d: asyncio.ensure_future(d.accept()))

Set viewport and user agent

ctx = await browser.new_context(
    viewport={"width": 1280, "height": 720},
    user_agent="Mozilla/5.0 ..."
)