@monoes/monobrowse

Browser automation via Chrome DevTools Protocol — navigate, click, fill forms, take screenshots, and evaluate JavaScript in a real Chrome browser. No Puppeteer, no Playwright, no external binaries — just CDP over WebSocket.

Part of the Monomind ecosystem. Powers monomind browse.

Install

npm install @monoes/monobrowse

CLI usage

# Via monomind
monomind browse open https://example.com
monomind browse screenshot --output page.png
monomind browse click "button.submit"
monomind browse eval "document.title"

Programmatic usage

import { Browser } from '@monoes/monobrowse';

const browser = new Browser();
await browser.connect();

await browser.navigate('https://example.com');
await browser.click('button.login');
await browser.fill('input[name="email"]', '[email protected]');
await browser.screenshot({ path: 'result.png' });

const title = await browser.evaluate('document.title');
await browser.close();

Features

• Zero dependencies on browser automation frameworks
• Direct CDP WebSocket connection to Chrome/Chromium
• Navigate, click, type, screenshot, evaluate JS
• Session management for multi-page workflows
• Headless and headed modes

Why not Puppeteer/Playwright?

Monobrowse talks CDP directly — no 50+ MB framework download, no binary management, no version pinning against Chrome releases. It does what AI agents need (navigate, interact, screenshot, evaluate) without the weight.

Links

• GitHub
• Documentation
• Issues

License

MIT