browser-cdp

name	browser-cdp
description	Native CDP browser automation reference. Headless/headed Chrome control, screenshots, JS evaluation. (/browser-cdp, browser automation, cdp, scraping)

Native Browser Automation (CDP)

Built-in browser control via Chrome DevTools Protocol — no Node.js, no Playwright, pure Rust.

Tools

Tool	Required	Optional	What
browser_navigate	url	headless	Navigate to URL
browser_click	selector	—	Click element by CSS selector
browser_type	text	selector	Type text into input
browser_screenshot	—	selector	Screenshot (returns file path)
browser_eval	script	—	Execute JavaScript
browser_content	—	selector, text_only	Extract text/HTML
browser_wait	—	selector, timeout_secs, delay_secs	Wait for element or delay
browser_find	—	pattern, mode, limit	Inventory (no pattern) or find elements

Headless vs Headed

• Headless (default): No visible window. Fast, low resources.
• Headed: Visible Chrome window. Pass headless: false to browser_navigate.

Usage Tips

• Compose workflows: navigate → inventory/find → click → type → screenshot
• Inventory mode: call browser_find with NO pattern to get every visible interactive element on the page, each with a stable [data-opencrabs-match="N"] selector. Prefer this over browser_screenshot when you have just landed and do not yet know what to click — text beats pixels for deciding where to act.
• Screenshots return file paths. Use <<IMG:path>> to send to channels.
• JavaScript evaluation for complex DOM manipulation and SPAs.
• No credentials stored. Navigate to login page and use click/type for auth (with user approval).
• Requires Chrome/Chromium installed. Feature-gated: browser flag.

Chrome Launch Failures

1. SingletonLock exists: rm -f ~/.opencrabs/chrome-profile/SingletonLock ~/.opencrabs/chrome-profile/SingletonCookie ~/.opencrabs/chrome-profile/SingletonSocket then retry.
2. DevTools data dir conflict: pkill -f chrome then retry.
3. Channel closed: Kill Chrome, delete lock files, retry.