May 16, 2026 · Free Guide

The Open-Source AI Browser That Passes Every Bot Detection Test

Every scraper hits the same wall. Cloudflare. reCAPTCHA. FingerprintJS. You try playwright-stealth, undetected-chromedriver, puppeteer-extra. They work for a week, then Chrome updates and they break again. CloakBrowser is different. It patches Chromium at the C++ source level, compiled into the binary. Detection sites see a real browser because it is a real browser. 12,000 GitHub stars in a few months. Free. Open source.

What this actually is

CloakBrowser is a custom-built Chromium binary with 49 fingerprint patches baked into the C++ source code. It ships as a Python pip package and an npm package.

You import it in place of Playwright. Same API. Same code. Stealth on by default. No JS injection, no flag-tweaking, no monthly subscription.

Why this beats every other stealth tool

Tools like playwright-stealth and undetected-chromedriver inject JavaScript to hide automation signals. Antibot systems detect those patches themselves. They also break every time Chrome updates.

CloakBrowser patches Chromium at the binary level, so there's nothing to detect. It's literally a real browser with a different fingerprint. Survives Chrome updates because it gets rebuilt with every major Chromium release (currently on 146).

The numbers that matter

  • reCAPTCHA v3 score: 0.9 (human-level, server-verified)
  • Cloudflare Turnstile: PASS (non-interactive and managed)
  • FingerprintJS: PASS
  • BrowserScan: NORMAL (4 of 4 checks passed)
  • deviceandbrowserinfo.com: 24/24 behavioral signals with humanize=True
  • Tested against 30+ detection sites
HOW TO INSTALL

One message to Claude. That's the install.

No build steps. No compiling. No config files.

Step 1. Open Claude Code in any folder.

Step 2. Paste this:

"Install CloakBrowser from https://github.com/CloakHQ/CloakBrowser and write me a starter script that opens a protected site and prints the page title."

Step 3. Claude pip-installs the package, downloads the stealth Chromium binary (~200MB, one-time), writes the script, runs it. You watch a real browser open and pass the detection. If you already have Playwright scripts, Claude will swap the import for you.

The 30-second test (no install needed)

Want to see it work before you install anything? Run this in your terminal:

docker run --rm cloakhq/cloakbrowser cloaktest

It spins up the stealth browser in a container, runs the full detection test suite, and shows you the pass/fail report. You'll watch Cloudflare Turnstile auto-resolve in front of you.

Support the project

Star CloakBrowser on GitHub

Free tools survive when people support them. This is genuinely the best stealth browser available right now and the maintainer is shipping updates weekly. Star it so it keeps getting built.

Star CloakBrowser on GitHub →

The one-line migration from Playwright

If you already have a Playwright scraper, this is the entire change:

- from playwright.sync_api import sync_playwright - pw = sync_playwright().start() - browser = pw.chromium.launch() + from cloakbrowser import launch + browser = launch()

Rest of your code works unchanged. page.goto(), page.click(), page.fill(), selectors. All identical.

The killer flag: humanize=True

Add one argument and CloakBrowser stops moving the mouse in straight lines and stops typing all characters in the same millisecond. Real Bézier curves. Per-character typing delays. Realistic scroll patterns.

browser = launch(humanize=True)

This is what beats behavioral detection. The thing that catches bots even when fingerprints are clean.

Browser Profile Manager (the Multilogin killer)

They also ship a self-hosted alternative to Multilogin, GoLogin, and AdsPower. Create unlimited browser profiles, each with unique fingerprints and proxies, persistent cookies and sessions.

Run it with one Docker command:

docker run -p 8080:8080 -v cloakprofiles:/data cloakhq/cloakbrowser-manager

Open localhost:8080 in a browser, click Launch, you're inside the stealth browser via noVNC. Multilogin charges $99/month for the equivalent.

When to use it. When to skip it.

Use it for: anything that gets blocked (e-commerce scraping, social media scraping, SERP scraping, ticket sites, lead generation), account creation flows, multi-account management, any automation that hits Cloudflare or reCAPTCHA.

Skip it for:scraping APIs that already work (don't add complexity you don't need), sites with no bot protection.

Worth knowing: CloakBrowser does not solve CAPTCHAs that DO appear. It prevents them from appearing in the first place. For high-volume work, bring your own proxies.

All Resources

CloakBrowser GitHub repoThe main repo. Star it.
CloakBrowser ManagerMultilogin alternative, self-hosted
PyPI Packagepip install cloakbrowser
npm Packagenpm install cloakbrowser
Docker HubPre-built container for the 30-second test

The Next Step

Guides show you what. AI Builders shows you how.

Inside the community, I walk through every build live — including the stuff that doesn't make it into guides. Regular people (not developers) figuring out AI together, shipping real projects, asking me anything. No fluff, no theory, just the actual work.

Join AI Builders

skool.com/ten-fold