feder-cr/invisible_playwright

Invisible Playwright is a Python web scraping and browser automation tool that extends Playwright with anti-detection capabilities to bypass bot detection systems.

The problem it solves is that standard browser automation tools are easily detected and blocked by anti-bot services like Akamai, DataDome, Imperva, PerimeterX, and Kasada, which then serve captchas. Invisible Playwright answers the two core questions anti-bots ask: it proves it is a real browser by patching Firefox at the C++ source level and setting the browser fingerprint inside the engine rather than injecting it into the page, covering navigator properties, screen dimensions, GPU and WebGL data, canvas, fonts, audio, WebRTC, timezone, and network details. It proves a real person is using it by humanizing driver actions, making every click, hover, and drag follow natural mouse paths with human timing and byte-identical input events rather than teleporting the cursor.

The tool maintains full API compatibility with Playwright, so existing code requires only two lines to switch. Each session gets a distinct fingerprint with randomized fields and Bezier-curve mouse motion, or you can pin specific fingerprint fields while deriving the rest from a seed for reproducibility. It supports socks5, socks4, http, and https proxies with DNS routed through the proxy by default, and allows timezone configuration. The tool requires Python 3.11 or newer and runs on Windows x86_64, Linux x86_64, and Linux arm64. A few Playwright surfaces are out of scope—tracing, HAR, CDP, and the API request context—and the tool explicitly refuses these with explanatory messages rather than failing silently.

The project shows consistent development activity with regular updates addressing edge cases and expanding anti-detection coverage. The codebase demonstrates careful attention to the technical details of fingerprinting, with documented constraints around GPU pinning that validate against a pool of known personas rather than accepting arbitrary values. Documentation includes guides on pinning specific fingerprint fields and how the Bayesian sampler interacts with manual overrides.