Browser Automation
検証済みControl a headless browser through natural language. Fill forms, scrape data, take screenshots, and automate repetitive web tasks. Built on Playwright with stealth mode to avoid bot detection. Perfect for price monitoring, form filling, and web testing.
インストール
$ claw install browser-automation 設定
環境変数は .env ファイルか skill manifest で設定する。
| VARIABLE | DESCRIPTION | DEFAULT |
|---|---|---|
| BROWSER_ENGINE | Browser engine to use (chromium, firefox, webkit) | chromium |
| HEADLESS | Run browser in headless mode | true |
| STEALTH_MODE | Enable anti-detection measures | true |
使い方
// Browser automation engaged
launch Launching headless Chromium...
check_circle Page loaded. Extracting pricing data...
Pro Annual: $24/mo (billed yearly)
Pro Enterprise: Custom pricing
Screenshot saved: ~/screenshots/pricing-2026-02-13.png
スキルマニフェスト（JSON）
{
"name": "browser-automation",
"version": "2.0.1",
"permissions": [
"browser.navigate",
"browser.interact",
"filesystem.screenshots"
],
"trigger": "regex:/(browse|scrape|fill|click|screenshot|open) .*(page|site|form|url|web).*/i",
"engine": "node-20"
}ユーザーの議論とフィードバック
Stealth mode is legit — no more CAPTCHAs on most sites. I use it to monitor 15 competitor pricing pages daily. Saves me hours of manual checking every week.
Replaced half my Selenium test suite with natural language commands. "Click the submit button and verify the success toast" just works. Cross-browser support is a nice bonus.
I have this government form I need to fill out monthly with the same data. Now I just say "fill the monthly report form" and it handles everything. Life-changing.