Summary of an article published on The Web Scraping Club on 2026-10-04 (free to read). Read the full article

Anti-bot systems look at two things: whether the browser is real and whether a real person is driving it. Most stealth tools for Playwright inject JavaScript or patch the driver, which leaves detectable surface. invisible_playwright, a Python library by Federico Elia, takes a different route. It patches Firefox in the C++ source and sets fingerprint data inside the engine, with the goal of being a drop-in replacement for Playwright.

I went through what the tool is and how it works: seed-based fingerprints, proxy handling, human-like input, the CLI, the MCP server and the framework integrations. Then we tested it. We ran it against four public bot detection pages, Fingerprint among them, and against a protected site for each of five major anti-bot systems, from my own residential IP. We also compared it with Chromium-based tools, playwright-stealth and Camoufox, and I give my view on where it fits and where it doesn’t.

What’s inside

  • How the single seed keeps GPU, canvas, fonts, audio and WebRTC consistent with each other, and why contradictions between signals give a spoofed browser away.
  • How pointer events are handled natively: timing, pressure, pointerType, movement deltas and sub-pixel coordinates, beyond plain Bézier mouse paths.
  • Why pinning individual fingerprint fields while the seed fills the rest is useful, and how it helps to replay a failed scrape.
  • The limits I see: Firefox only, Python only, a single main maintainer, and the risk that anti-bot vendors reverse-engineer an open-source project as it spreads.

In the full article: Code samples for seeds, pins, proxies and the MCP server, the setup steps, the screenshots of each detection test, the benchmark table per anti-bot system and the comparison table with other stealth tools.

Read the full article on The Web Scraping Club