Patchright
Measured against stock Playwright: no difference (in this mode).
Patchright is a drop-in Playwright fork with anti-detection patches. The pitch is simple — same API, fewer blocks — so we tested it the same way we tested stock Playwright: version 1.60.0, headless Chromium, no proxy, 21 sites, same day.
Result: identical. The same 3 passes (Target, Rightmove, Wikipedia), the same 18 failures with the same reasons, at near-identical latencies. Amazon still 503, the 403 wall still 403s, the captcha sites still serve captchas, Reddit still ships an empty shell. On this corpus, headless and unproxied, the patches changed nothing we could measure.
That is a mode-specific finding, not a verdict on the whole idea: modern detectors weigh IP reputation and headless signals heavily, and both were held constant and unfavorable here. Patchright headed, or behind residential proxies, might separate from stock — we haven't tested those modes yet. But the drop-in-stealth claim, taken at face value, did not survive contact.
Test conditions
21 real public sites (public-tough-v1), 2026-05-24, no proxy, single residential IP, Linux. Deterministic validators: expected status code plus required and forbidden content markers. Pass rates measure the tool and the network position together. Without a proxy, results reflect how each site's protections treat one plain residential connection as much as the tool itself.
What worked
- +Fully API-compatible: our Playwright adapter ran unmodified.
- +Rendering, artifacts, and stability match stock Playwright exactly.
What failed
- −Zero measured improvement over stock Playwright: same 3/21, same failure reasons, same latencies.
- −The default configuration most people will try first — headless, no proxy — is precisely the mode where it showed no benefit.
When to use it
If you're already invested in Playwright and want a free lottery ticket against detectors, it costs nothing to swap in. Don't expect headless-without-proxy miracles; test your specific targets before believing any stealth claim, including this one.
In the mode everyone actually uses first, it performed exactly like the thing it patches.
Run evidence
Every attempt below is backed by saved artifacts: request/response metadata, body content, and (for browsers) screenshots. Raw JSON and artifacts live in the runner data directory.
Run data not available in this build.