Keyboard test results apply to the sampled keys and combinations. Compare isolated presses, complete releases and held groups separately. Use the keyboard ghosting guide when several keys fail together, rather than treating a completed single-key map as proof of every combination.
Keyboard test
Use the keyboard test to compare key registration, releases and held combinations. Keep automatic repeat separate from a new press after release, and record the event sequence before assigning a hardware cause.
Also known as: keyboard test, test keyboard, keyboard tester, key tester, key test, what key am i pressing. Free, online, and no download required.
No permission neededNothing uploaded
Press any physical key to begin the keyboard run.
01
What the keyboard test checks
| Check | How it is measured | Fails when |
|---|---|---|
| Key registration | KeyboardEvent.code maps each physical position independently of the printed character. | A physical key never produces a keydown event. |
| Key release | Every keydown is paired with keyup and monitored for three seconds. | A key stays down after its physical release. |
| Rollover | The largest simultaneous held-key set is recorded during the run. | A combination drops or replaces a held physical position. |
How to run the keyboard test
- Choose the layout that matches your board. Keep the keyboard, connection and browser fixed, then click an empty part of the instrument so a search box, menu or form is not receiving your keys.
- Select Reset coverage before a fresh sample. Press A once and check down KeyA in Event log, the highlighted A position and Held now. Release A fully and check up KeyA and Held now returning to 0.
- Repeat isolated presses and full releases for the keys you need. Proven keys counts distinct received positions; a proven position is not a certificate that its switch or release works in every application.
- Select Run rollover check to clear the current coverage and log. Hold A, add S, then add D without releasing the earlier keys. Compare Held now and Maximum held with the three intended keys, then release all three and check Held now is 0.
- Read the latest five Event log entries immediately after each short sequence. Automatic repeat can add down entries while a key is held; it is not a new physical tap. Repeat a missing release without switching windows before assigning a cause.
- Select Finish and read result, record the sampled keys, press order and verdict, then Reset coverage before a comparison. Change only one condition. Avoid system-reserved shortcuts; Block browser shortcuts cannot capture keys the operating system intercepts.
If something failed
Use the same keyboard, layout, connection, application and press order. Reset between comparisons and fully release each isolated press. Record intentional holds separately from fresh taps. If a combination fails, change only one group or order per run so you can describe what changed without mixing several tests.
The browser cannot inspect contacts, matrix wiring, diodes, scan rate or USB reports. System shortcuts and remapping can change which events arrive. A key tester observes the delivered sequence, not the full hardware route.
02
How to read the result
What key am I pressing?
Press one key. The keyboard tester places the received KeyboardEvent.code on the map, separately from the character printed by the active layout. A remapped or virtual input can change what reaches the browser. The key tester cannot identify the physical switch by inspecting its label.
Does the position stay highlighted after release? Clear the run and repeat without switching windows. A missing keyup can follow a focus change or intercepted event. The map records an unmatched down event; it does not establish whether the switch failed to return physically.
For the question what key am i pressing, compare the code with the key you intentionally used. Choose the layout for the visible board, then check the event log. A layout label and a received character answer different questions, so do not call a position dead merely because its printed letter differs.
Keyboard double typing and automatic repeat
Hold, then release. A held key can produce keydown events with repeat=true; that is automatic repetition, not a series of fresh physical presses detected by this page. For keyboard double typing, compare isolated presses with full releases rather than holding the key to generate extra events.
Does a released key create another action? Record the down and up order and compare the affected application. Short intervals alone cannot distinguish a switch, remapping profile or software interpretation. This key test does not measure electrical bounce or certify the cause of repeated characters.
Use the keyboard repeat rate test for held-key timing and the keyboard check guide for missing or repeated input. To test keyboard combinations, add ordinary keys one at a time and compare another press order. A missing combination deserves a separate rollover observation.
How the keyboard test works
The map uses event.code for received positions and tracks down, up, held and proven states. The latest event entries support a short sequence comparison, including key repeat. A three-second unmatched-down threshold is this page's review rule, not proof of a physical sticky switch. Follow the how to test keyboard guide for a controlled comparison.
Keep the same keyboard, layout, connection, foreground browser and press order between runs. Inputs directed at search, forms, navigation or dialogs are excluded. The map distinguishes untested, proven, held and stuck positions; Event log shows only the latest five down or up entries, not a saved transcript.
What the result can establish
Keyboard test results apply to the sampled keys and combinations. Compare isolated presses, complete releases and held groups separately. Use the keyboard ghosting guide when several keys fail together, rather than treating a completed single-key map as proof of every combination.
Proven keys counts distinct positions that delivered a non-repeat down event. Held now is the currently unmatched down set; Maximum held is the largest set seen since reset. Stuck flags a position still unmatched after three seconds, including an intentional long hold. This is not proof of a physical sticky switch; fully release and repeat while the tab stays focused.
Finish and read result needs at least ten delivered events for a verdict. Fewer events are inconclusive, and a clean sample does not guarantee hardware health. Key repeat rate, Median and p95 may stay blank on this general test; use the dedicated keyboard repeat rate test for held-key timing.
Reset coverage and Run rollover check both clear the current map, held set, maximum and recent log. Copy the short sequence into your own notes before either action. Repeat a failed combination in another order, keeping its intended keys explicit.
Measurement limits
The browser cannot inspect contacts, matrix wiring, diodes, scan rate or USB reports. System shortcuts and remapping can change which events arrive. A key tester observes the delivered sequence, not the full hardware route.
Cmd+Tab, Alt+Tab, Ctrl+Alt+Del and PrintScreen may be reserved before the browser receives them. F5, F12 and Cmd+R remain available even with Block browser shortcuts enabled. A missing reserved action is not a hardware failure, and a focus change can leave a release unobserved. Compare the original symptom in its actual application.
On touch or coarse-pointer devices, this page keeps the explanation but shows a desktop-input requirement instead of mounting the keyboard instrument. A narrow desktop browser window can still receive keyboard input; viewport width alone is not a physical-device test.
Content review and test scope
Content reviewed on 2026-10-10 by ProbeKits Team for correspondence with the browser controls and readouts; this is not a professional hardware inspection.
Physical-device results: not measured. Automated key events verify software interaction only, not switch condition or real-device rollover. For your own observation, record the keyboard model, browser/version, layout, connection, keys and order, delivered log and what remains unknown.
03
Keyboard test questions
Can I test a keyboard online without typing into a document?
Yes. The keyboard test reads events delivered to this page, so a document is not needed. Focus the test rather than a search or form field, press and release the keys, then inspect the map. System-reserved shortcuts may still act outside the page and should not be used as hardware pass criteria.
Why does this key tester use physical positions, not letters?
KeyboardEvent.code describes the received key position independently of the character chosen by the active layout. That helps compare positions across layouts. Remapping and virtual input can still change the delivered event. Compare the key you pressed with the visible code; this page cannot inspect the switch or its printed label directly.
A key stays highlighted after I release it. Is it broken?
Not necessarily. The page may have missed a release after a window or focus change, or the input may remain active in the delivered stream. Clear the run and repeat a normal press and release without changing focus. Record a repeated unmatched event before attributing it to a physical return problem.
What key am I pressing right now?
Look at the code in the recent event log and the matching highlighted position. The map reflects input received by the browser, not a photograph of your board. If the label differs from the printed character, compare the active layout and mapping before treating the position as a failed key.
How do I test keyboard rollover here?
Hold one ordinary key, add the next without releasing it, and continue through the combination you need. Record the largest held set and which input is absent. Repeat the group in another order and region. A missing event demonstrates the browser observation, but cannot identify a matrix or USB protocol cause.
How should I repeat the keyboard test fairly?
Use the same keyboard, layout, connection, application and press order. Reset between comparisons and fully release each isolated press. Record intentional holds separately from fresh taps. If a combination fails, change only one group or order per run so you can describe what changed without mixing several tests.
What can the keyboard test not prove?
It cannot prove switch health, electrical bounce, matrix design or the device report rate. It also cannot inspect another application's bindings. The keyboard test confirms received events during a sample; use repeated observations and the keyboard guide to choose a relevant next comparison.