inputprobe Keyboard Tester's answer

It's the keyboard tester that refuses to guess. Most testers ship a "chatter detected!" verdict based on an invented millisecond threshold, but nobody has published where deliberate double-tapping ends and switch flutter begins, and a guessed threshold is how tools end up accusing healthy hardware. This tester shows you the evidence instead: per-key press counts, auto-repeat filtered out, live and maximum simultaneous-key counts (rollover), and each key's fastest re-press time, and leaves the verdict with you, with guidance on how to judge it against your own fastest deliberate double-tap. It's also honest about what a browser can and cannot see: it documents which keys legitimately never reach a web page (laptop Fn, media keys, system shortcuts) so you don't blame working hardware, and it states plainly that rollover is measured as delivered through the OS, not at the keyboard's connector.

inputprobe Keyboard Tester's answer

Because it won't lie to you in either direction. It never accuses a healthy key with a made-up chatter threshold, and it never clears a dead one, a key that doesn't light gets explained (switch, connection, or a layer below the browser), with the legitimately-invisible keys documented so silence isn't misread as failure. Practical details are handled properly: keys are captured before the browser acts, so Space doesn't scroll and Tab doesn't leave the page, with a Pause button that hands your keyboard back; the board runs on physical key codes, so it works on any layout, not just US QWERTY; and macOS's quirk of not reporting a letter key released while Command is held is corrected rather than counted against the key. Everything runs locally, nothing you type leaves your machine.

inputprobe Keyboard Tester's answer

Mechanical keyboard owners diagnosing a switch that types twice (and deciding whether it's worth a switch swap), anyone checking a used or secondhand laptop's keyboard before buying, gamers verifying that their board actually delivers the rollover it advertises, and people troubleshooting a single dead or flaky key who want to know whether it's the switch, the connection, or the operating system eating the keypress.

inputprobe Keyboard Tester's answer

It's part of inputprobe, a set of input-device testers built around one rule: a verdict must quote the evidence it's based on, and when the evidence isn't good enough, the tool refuses to judge. For the keyboard tester that rule had a concrete consequence: every other tester publishes a chatter verdict from a threshold nobody has measured, and we haven't yet measured where human double-tapping ends and switch flutter begins on real keyboards, so this page deliberately ships no verdict, displays the timing data, and says exactly why. The reasoning is written up at inputprobe.com/methodology/, alongside what has and hasn't been validated on real hardware across the whole site.

inputprobe Keyboard Tester's answer

Plain web platform, no frameworks: dependency-free JavaScript ES modules reading KeyboardEvent physical key codes, with events captured ahead of browser default actions so testing doesn't scroll or navigate the page. The site is static HTML generated by a small Python build (Jinja2 + Markdown) and served from Cloudflare's edge. There is no backend, every keystroke is read, counted, and displayed entirely in your browser, and nothing is uploaded.

inputprobe Keyboard Tester's answer

None to list. It's a free tool with no accounts and no tracking, so we genuinely don't know who uses it, and we prefer it that way.