[BUG] There appear to be invalid fingerprints in the apify dataset that are leading to failures during runs, especially with the new FF146 version of camoufox

[JWriter20]

These fingerprints create "voids" that playwright thinks are elements in, but these are not actually clickable. Click actions fail.
I think the 135 Firefox version handled these somehow, as I really didn't notice these issues, but a bunch of my tests became flaky and occasionally failing because of this in the newer 146 version.

Example invalid fingerprint, inner width and height are greater than the outer width and height. This makes no sense
{
"screen": {
"availTop": 0,
"availLeft": 0,
"pageXOffset": 0,
"pageYOffset": 0,
"screenX": 772,
"hasHDR": false,
"width": 1536,
"height": 960,
"availWidth": 1536,
"availHeight": 960,
"clientWidth": 0,
"clientHeight": 21,
"innerWidth": 1280,
"innerHeight": 720,
"outerWidth": 751,
"outerHeight": 886,
"colorDepth": 24,
"pixelDepth": 24,
"devicePixelRatio": 1
},
"navigator": {
"userAgent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:135.0) Gecko/20100101 Firefox/135.0",
"userAgentData": {
"brands": [],
"mobile": false,
"platform": "Windows",
"architecture": "x86",
"bitness": "64",
"fullVersionList": [],
"model": "",
"platformVersion": "10.0.0",
"uaFullVersion": "135.0"
},
"doNotTrack": "unspecified",
"appCodeName": "Mozilla",
"appName": "Netscape",
"appVersion": "5.0 (Windows)",
"oscpu": "Windows NT 10.0; Win64; x64",
"webdriver": "false",
"language": "en-US",
"languages": ["en-US", "en"],
"platform": "Win32",
"hardwareConcurrency": 8,
"product": "Gecko",
"productSub": "20100101",
"vendor": "",
"vendorSub": "",
"extraProperties": {
"vendorFlavors": [],
"isBluetoothSupported": false,
"globalPrivacyControl": null,
"pdfViewerEnabled": true,
"installedApps": []
}
},
"videoCodecs": {},
"audioCodecs": {},
"pluginsData": {},
"videoCard": {
"renderer": "ANGLE (NVIDIA, NVIDIA GeForce RTX 3060 Direct3D11 vs_5_0 ps_5_0, D3D11)",
"vendor": "Google Inc. (NVIDIA)"
},
"multimediaDevices": [],
"fonts": [],
"mockWebRTC": false
}

[barjin]

Hello, and thank you for your interest in this project!

outerWidth reports the actual width of the browser window on your screen, innerWidth is the viewport size (i.e., how many pixels of content can the browser show). The difference is based on your zoom level (check your innerWidth with different zoom levels and see the difference).

These values in the fingerprint therefore might make sense. The other behaviour you describe, however, does not. Can you please expand on the ""voids" that playwright thinks are elements in, but these are not actually clickable"? Can you try replicating the issue with a script (and share this script here)? Thank you!

[JWriter20]

I have attached a reproduction script and videos of what a successful vs bad run looks like.

Bug Reproduction Video

[Bug reproduction vid](https://github.com/user-attachments/assets/404d9593-ee58-43cb-9360-1ea8f5c9040b)

---

Issue Description

All click actions are timing out.

Example logs:

Click failed (cursor.click timed out after 8000ms for #resumator-firstname-value) for #resumator-firstname-value, scrolling into view and retrying once
Retry after scroll failed for #resumator-firstname-value: cursor.click timed out after 8000ms for #resumator-firstname-value
click failed, NOT falling back to native click (useCursor=true): #resumator-firstname-value
cursor.click timed out after 8000ms for #resumator-firstname-value
Error clearing field: Error: cursor.click timed out after 8000ms for #resumator-firstname-value

Clean Run (for comparison)

[Clean run video](https://github.com/user-attachments/assets/347349d0-4d10-442b-b414-f1a380875206)

---

Additional Observations

In the buggy run, you can see the "void" at the bottom of the page. I have also had this void appear on the right side of the page and nearly take up the entire size of the active page content.

The mouse tries to move over to the fields but cannot reach them. I am fairly confident this is not just a consequence of the fixture itself — during resizing the fields are still always present. It seems much more likely to be an issue with Camoufox or camoufox-js.

---

Reproduction

It should be as simple as running the following script with Camoufox 146:

import { Camoufox } from 'camoufox-js';

const fingerprint = {
    screen: {
        availTop: 0,
        availLeft: 0,
        pageXOffset: 0,
        pageYOffset: 0,
        screenX: 772,
        hasHDR: false,
        width: 1536,
        height: 960,
        availWidth: 1536,
        availHeight: 960,
        clientWidth: 0,
        clientHeight: 21,
        innerWidth: 1280,
        innerHeight: 720,
        outerWidth: 751,
        outerHeight: 886,
        colorDepth: 24,
        pixelDepth: 24,
        devicePixelRatio: 1,
    },
    navigator: {
        userAgent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:135.0) Gecko/20100101 Firefox/135.0',
        userAgentData: {
            brands: [],
            mobile: false,
            platform: 'Windows',
            architecture: 'x86',
            bitness: '64',
            fullVersionList: [],
            model: '',
            platformVersion: '10.0.0',
            uaFullVersion: '135.0',
        },
        doNotTrack: 'unspecified',
        appCodeName: 'Mozilla',
        appName: 'Netscape',
        appVersion: '5.0 (Windows)',
        oscpu: 'Windows NT 10.0; Win64; x64',
        webdriver: 'false',
        language: 'en-US',
        languages: ['en-US', 'en'],
        platform: 'Win32',
        hardwareConcurrency: 8,
        product: 'Gecko',
        productSub: '20100101',
        vendor: '',
        vendorSub: '',
        extraProperties: {
            vendorFlavors: [],
            isBluetoothSupported: false,
            globalPrivacyControl: null,
            pdfViewerEnabled: true,
            installedApps: [],
        },
    },
    videoCodecs: {},
    audioCodecs: {},
    pluginsData: {},
    videoCard: {
        renderer: 'ANGLE (NVIDIA, NVIDIA GeForce RTX 3060 Direct3D11 vs_5_0 ps_5_0, D3D11)',
        vendor: 'Google Inc. (NVIDIA)',
    },
    multimediaDevices: [],
    fonts: [],
    mockWebRTC: false,
};

async function main() {
    const browser = await Camoufox({
        headless: false,
        fingerprint,
        i_know_what_im_doing: true,
    });

    const context = await browser.newContext({ viewport: null });
    const page = await context.newPage();

    await page.goto('https://example.com');
    console.log('Loaded example.com — sleeping 30s so you can inspect.');
    await page.waitForTimeout(30_000);

    await browser.close();
}

main().catch((e) => {
    console.error(e);
    process.exit(1);
});

---

Versions Tested

• My custom build (with minor unrelated fixes, pending merge into main): [Camoufox 146.0.1-beta.25.2](https://github.com/JWriter20/camoufox/releases)
  Built from main source with 2 minor fixes unrelated to this

Would likely reproduce on the release too: https://github.com/daijro/camoufox/releases

Platforms Affected

Reproduces on both macOS and Linux.

[barjin]

@JWriter20 thank you for the detailed write-up on this issue.

Afaiac, the window scaling artifacts (the "void") have always been present for managed browsers with injected fingerprints. I always treated them as a purely visual issue.

As you mention in the initial comment,

I think the 135 Firefox version handled these somehow, as I really didn't notice these issues, but a bunch of my tests became flaky and occasionally failing because of this in the newer 146 version.

This is leading me to think that the issue is in the Camoufox binary and not the JS wrapper.

Can you please try reproducing this issue with the original Python Camoufox project? If it reproduces, I'm afraid there is not much we can do here.

Thanks again!