How We Test

ScreenTail is a set of browser tools for testing and using screens, run by Vivek Kumar. This page describes how they are checked before anything is published, and keeps three things apart: the automated checks, the way the work is done, and Vivek's own checks on real screens.

Read what no browser test can do first if you are deciding whether to trust the rest.

There is no account, no upload and no server doing any of the work. Every page is a static file, every tool is code running on your own device, and nothing about your screen is sent anywhere. ScreenTail sets no cookies, and the only message a page sends is one page-view count with the page's address. You do not have to take any of that on trust: open your browser's network panel and watch what a tool page requests.

What the site stores, and anything it might add later, is set out in the Privacy Policy.

Every change runs the same chain of checks, and nothing is published unless all of them pass. At the time of writing that is the build and type checks, a data check, a link check, and more than 450 checks in real browsers.

• The site builds with no type errors and no warnings.
• The tool data matches the address list, so a page cannot appear at an address it was not meant to have.
• Every internal link is followed and must lead to a published page.
• Every page is opened in a real browser, in the light and dark themes, at six widths from a wide desktop down to a phone. Nothing may spill past the edge, and the browser's console must report no errors.
• The tools are used, not just loaded. Buttons are pressed, keys are sent, full screen is entered and left, and patterns are read back pixel by pixel.
• Nothing is left running. After a page is left, the checks count the timers, listeners and animation frames still alive. The number has to be zero.
• The public site holds only published pages. Before anything goes live, the files to be published are checked so that no draft, unfinished page or placeholder can be among them.

A check that always passes is worth nothing. The checks that guard safety and privacy were shown to fail first: the brightness check was fed a deliberately flashing sequence of numbers (never drawn on a screen), and the check on the public files was given planted drafts, placeholders and broken links, and caught every one.

They have also caught real faults before publication. A few of them, as recorded while the tools were built:

• The Flip Clock's flip first changed brightness too fast; it was slowed to a steady 1.2 seconds.
• The Lava Lamp drew only 55 frames in 3 seconds in full screen; it now draws smoothly.
• The browser's own gradient added noise to gradient wallpapers, which the page promised not to do; every pixel is now worked out exactly.
• On the Broken Screen, a click first added nothing; it was fixed before release.

Building and routine testing follow a written, step-by-step workflow. Each tool is made against a written specification, runs through every automated check above, and is recorded, including the faults found on the way, before it can be published. Screenshots of every page, at desktop and phone widths and in both themes, are reviewed as part of that workflow.

That review is a useful second look, but it is not a person looking at a real screen, which is what the next section is for.

Automated checks prove that a page works. They cannot tell you whether a gray looks even on a real panel or whether a test is easy to follow. So Vivek has checked the website himself on three kinds of screen: a phone, a laptop and a 4K monitor. It works as it should on each of them.

That is one person with three screens, not a laboratory. A fault that shows only on your particular hardware can still slip through, which is why a report from you is so useful.

Every figure and every manufacturer policy in a guide is checked against a named source, with the date it was read, before the guide is published. Where no source could be found, the guide says so rather than inventing a number. Each guide lists its sources at the end, and each is signed by Vivek Kumar.

No guide or tool makes a health or medical claim.

ScreenTail is run by one person, named, and accountable for what is published. If a tool behaves oddly on your screen, write to contact@screentail.com (the contact page says what helps) with the tool, your device and your browser. Reports like that are the only way some faults are ever found.

About ScreenTail explains what the site is and who runs it.