Device

Keyboard test: check received keys, combinations and repeat counts

·

Select Start keyboard test, then press keys inside the focused test area. ToolMellow highlights received physical codes, shows the latest key event and keeps per-code press and repeat counts. You can test combinations and copy or download the accumulated results.

A browser keyboard test checks events that reach this page. It does not certify every physical switch, measure input latency or prove hardware rollover and ghosting behavior. Use the reported values to describe what you observed, while allowing for focus, keyboard layout and shortcuts reserved by the browser or operating system.

Keyboard test

Start inside the focused test area

Open the keyboard tester and select Start keyboard test. The test area receives focus automatically. Press a letter, release it and check that the corresponding physical-code position changes appearance. Opening the page alone does not start recording key results.

Try individual keys before holding a small combination. A currently held code has a different highlight from a code received earlier. The diagram is a visual aid; the received-keys table and exported report provide the actual code, key, location and counts.

Keep the test area focused while testing. On a laptop, try the letter row, modifiers and navigation keys you use regularly. If you have a numeric keypad, inspect its received codes and locations as well as the visible highlights.

W3C UI Events — keyboard events, location, repeat and modifier statesKeyboardTester — keyboard tester and numeric keypad notes

Choose a diagram without changing the OS layout

The Keyboard layout menu offers Full ANSI with numpad, Compact ANSI and Full ISO extra key. Compact hides the separate navigation and numpad groups. ISO adds the physical code IntlBackslash beside the left Shift position; it does not add a language-specific character map.

Changing the diagram stops the active test and clears the currently held highlight, while retaining received-key history, counts, maximum and last observed lock values. It does not change the keyboard layout configured in your operating system. Previously received codes remain in the report even when the selected diagram does not draw them.

W3C UI Events KeyboardEvent code values — physical positions

Read the values rather than the printed keycap

The browser’s key value describes the event’s meaning under the current layout and modifiers. Its code identifies a physical position where one is reported. For example, the position reported as KeyQ can produce a character other than q on another language layout.

Rows are grouped by code. A later event for the same code updates its key and location while retaining aggregate press and repeat counts. The table is not a chronological transcript, and a label on the diagram is not a prediction of the character the browser will report.

Read the values rather than the printed keycap
ValueWhat it describesHow to use it
KeyLatest browser-reported event meaningCompare it with the selected OS language layout
CodeReported physical positionUse it to distinguish positions with similar labels
LocationBrowser-reported location numberSeparate left, right or numpad events where available
PressesKeydown events not marked as repeatsAn aggregate count for that reported code
RepeatsKeydown events marked as repeatedSeparate hold-repeat behavior from new presses
Distinct codesNumber of received code recordsNot the number of verified physical switches
Held nowCodes currently tracked as heldClears when capture stops
Maximum togetherLargest observed held-code setAn observation from this session, not an NKRO rating

W3C UI Events — keyboard events, location, repeat and modifier statesW3C UI Events KeyboardEvent code values — physical positions

Interpret location and missing-code fallbacks

Location values help distinguish left and right modifiers and the numeric keypad. Look at both code and location: ShiftLeft and ShiftRight identify different positions, while Numpad1 is distinct from the digit-row key even when they can produce the same character.

If the browser supplies no code, the tool groups that event using a fallback based on its location, such as Unidentified:0. Different unreported keys sharing that location can therefore merge into one row. Missing or unusual browser data is not a reliable count of separate physical keys.

Interpret location and missing-code fallbacks
Value or fieldMeaningBoundary
Location 0Standard or unspecified locationDoes not identify a particular keyboard model
Location 1Left-side locationTypically used for left modifiers
Location 2Right-side locationTypically used for right modifiers
Location 3Numpad locationDepends on the browser event actually received
Empty codeFallback grouped by reported locationDifferent keys can share one fallback row
Interface languageLanguage of the page and its number formattingDoes not select the physical or OS language layout

W3C UI Events — keyboard events, location, repeat and modifier states

Separate held keys from received history

While the test is active, keydown adds the reported code to the held set and keyup removes it. Holding two received codes together can show two held positions. Repeated keydown events for one code do not create additional held positions.

After a key is released, its received-history highlight remains. Stop test or a focus/lifecycle stop clears current held state but keeps the accumulated results. Use Reset highlights when you want a clean result set rather than treating an old highlight as a key that is still down.

Use the simultaneous count as a local observation

Maximum together is the largest number of distinct codes the page has tracked as held during the accumulated test. Try the combination you care about, release it and inspect the received codes. A missing key event can have several causes beyond the keyboard’s switch matrix.

This is not a controlled NKRO or ghosting certification. Browser focus, OS-reserved combinations, missing keyup events, virtual inputs and unreported codes can affect the observation. The tool also has no polling-rate, debounce, switch-bounce or end-to-end latency measurement.

Compare new presses with hold-repeat events

A keydown with repeat false increments Presses; one marked true increments Repeats. Holding a key can therefore add repeats without adding another press. The tool follows the browser flag and does not infer switch chatter from a sequence of events.

Repeat delay and rate depend on system settings and browser behavior. The report stores aggregate counts, not event timestamps, so it cannot calculate milliseconds between presses or a repeat frequency in hertz. Compare counts only in the context of how you actually tested.

W3C UI Events — keyboard events, location, repeat and modifier states

Read lock labels as the last observation

The CapsLock, NumLock and ScrollLock pills begin as Not observed. Each recorded keydown refreshes their values from the browser’s modifier-state report. A displayed value can remain after you stop the test.

Off / unsupported combines a false state with the possibility that the browser or platform does not support that observation. It is not a separate hardware LED test. Start a focused test and press a suitable key to obtain a new observation; Reset highlights returns the lock display to its initial state.

Allow for reserved shortcuts and composed input

Browser and operating-system shortcuts can be handled before the page receives them. Fn and device-specific media or hardware keys may also be handled internally. A missing event is not enough to conclude that the key is broken, and the tester does not offer a keyboard-lock mode that captures every reserved combination.

The page can receive values such as Dead or Process during dead-key or input-method handling. It does not record composition or text-input events separately, or export a composition flag. Ordinary active keydown defaults are canceled, which can also affect text composition; use a normal text editor to check the final composed characters.

W3C UI Events KeyboardEvent key values — meanings and special keys

Use Tab to leave the test safely

Tab is deliberately allowed to move focus out of the test area, which stops capture. The Tab event itself can be included in the accumulated results. The test also stops when its area or the window loses focus, the document becomes hidden or the page is being left.

While active, ordinary non-Tab keydown events without Ctrl, Alt or Meta have their browser default prevented. Combinations with those modifiers retain their defaults, so an action such as a browser shortcut may still happen. Start keyboard test again to resume after focus has moved; typing elsewhere on the page is not added to the test.

WHATWG HTML — focus and keyboard interaction

Choose between stopping and clearing results

Stop test ends active capture and releases the held highlights while preserving accumulated records and the maximum. Reset highlights stops capture and clears received keys, counts, maximum, last lock values and the latest-event detail. Starting again after a stop continues the retained result set unless you reset first.

Copy report and Download report stop active capture before exporting the retained data, and the chosen button receives focus. Leaving the tool or reloading creates a fresh instance rather than saving a report library inside the page.

Choose between stopping and clearing results
ActionCapture and held stateAccumulated results
Open the pageIdle; nothing captured yetNo received-key records
Start keyboard testArea receives focus; capture activeRetained results continue if not reset
Stop testCapture stops; held set clearsRecords, counts, maximum and last locks remain
Tab or focus lossCapture stops after leaving the areaEarlier results remain; Tab may be included
Hidden document or page leavingActive capture stopsNo persistent report library is created
Change diagramCapture stops; held set clearsExisting records remain under the new selection
Reset highlightsCapture stops; held set clearsRecords, counts, maximum, locks and detail clear

Inspect a report before sharing it

Copy report asks the browser clipboard API to write the current JSON. If clipboard access fails, the page suggests downloading instead. Download report saves toolmellow-keyboard.json. Both actions become available once at least one code has been recorded.

The JSON includes the selected diagram, page language, code counts, largest observed held set, last lock values and aggregate key records. Each record contains the latest key value and location plus press and repeat counts. It includes character values typed in the test area, so inspect it before sharing; it is not an anonymized hardware certificate.

Inspect a report before sharing it
JSON fieldSaved valueInterpretation
layoutfull, compact or isoSelected visual diagram rather than OS language layout
interfaceLanguagePage language tagExplains interface language and formatted labels
distinctCodes and maxSimultaneousCode-record count and largest held setAggregate observations for the current result set
lastReportedLocksLatest lock observation or nullNot a continuously refreshed hardware reading
keysPer-code latest key/location and press/repeat countsNot a timestamped raw-event timeline
NoteLocalized explanation of browser-event limitsDoes not add missing events or hardware evidence

Keep test data and exported files separate

The keyboard module processes events in page memory and has no automatic report upload or persistent-storage path. It only records active test-area keydown events. This does not mean the site never loads ordinary page resources; it means typing elsewhere is not a keyboard-test result.

Avoid typing passwords or other sensitive text into the active area because key values can reveal characters in the table and report. Stop ends capture; Reset highlights clears this page’s results. Copied clipboard content and downloaded files are separate copies, and resetting the page cannot erase them.

Check focus and event meaning before diagnosing hardware

If nothing appears, select Start keyboard test and confirm that the test area has focus. If a printed character differs from the result, compare key, code, location and the OS input layout. A key can be received and listed even when its code has no drawn position in the current diagram.

If a combination is incomplete, first test each key on its own, then the combination, and note which codes the browser actually receives. Check for reserved shortcuts and focus changes. When investigating a suspected physical problem, compare behavior in another ordinary application or a controlled hardware test rather than treating one missing browser event as a final diagnosis.

Make a small, repeatable check

Reset between comparisons when you need separate result sets. Keep the browser, OS input layout and tested combination consistent, and write down what you pressed alongside the downloaded report. A fresh baseline makes the accumulated counts easier to interpret.

Use the table and report to describe observed events accurately. A successful local test can narrow a problem, but it does not verify another application, an operating-system shortcut path, every physical switch or a future session with different input settings.

  • Choose the diagram that is useful for the physical positions you want to inspect.
  • Start the test and check that its area has focus before pressing keys.
  • Press and release individual keys before trying simultaneous combinations.
  • Compare key, code, location, presses and repeats instead of relying only on the diagram.
  • Stop or use Tab to leave capture; reset when you need to clear accumulated results.
  • Inspect a copied or downloaded report before sharing its character values.

W3C UI Events — keyboard events, location, repeat and modifier states

Frequently asked questions

Why do some keyboard keys not appear?

The page only sees events delivered to its active focused area. OS or browser shortcuts, Fn handling, missing browser codes and a code absent from the selected diagram can explain a missing highlight. Inspect the received-keys table before treating the result as a hardware fault.

What is the difference between key and code?

Key describes the event’s meaning under the current layout and modifiers; code identifies a physical position where reported. One position can produce different characters across input layouts, so the tool retains both rather than translating the diagram into a language map.

Why does pressing Tab stop the keyboard test?

Tab keeps its normal focus-navigation behavior. Leaving the test area stops capture and clears held state while retaining accumulated results. The Tab keydown itself may be recorded before focus moves. Select Start keyboard test to continue.

How are simultaneous keys counted?

The page adds distinct reported codes on keydown and removes them on keyup. Maximum together is the largest held-code set observed before a reset. Repeat events for the same code do not add another held position, and this count is not a hardware rollover certification.

How do repeat events differ from key presses?

A keydown marked as a repeat increments Repeats instead of Presses. System settings influence repeat behavior. The JSON has aggregate counts without timestamps, so the tester cannot calculate repeat rate, polling rate or input latency.

What do the lock-state labels mean?

They show the last browser-reported CapsLock, NumLock and ScrollLock values from a recorded keydown. Not observed means no value has been recorded yet; Off / unsupported does not distinguish a false lock state from unavailable platform support.

Can this keyboard tester detect Fn and every shortcut?

No. Fn and device-specific keys may be handled internally, while the browser or OS can reserve shortcuts before the page receives them. The tool preserves normal Tab and Ctrl/Alt/Meta shortcut defaults and does not claim to capture every possible combination.

What is saved in the keyboard report?

It saves the selected diagram, interface language, distinct-code count, largest observed simultaneous set, last reported locks, and per-code latest key/location with press and repeat counts. Character values from the test area are included; timestamps and text typed outside that area are not.

Does this keyboard test verify NKRO or ghosting?

No. It reports browser-received codes and their observed held set. Missing events, focus, OS shortcuts, fallback-code grouping and the input environment can affect the count. Controlled hardware testing is needed for a physical NKRO or ghosting conclusion.

Does selecting ANSI or ISO change my keyboard language?

No. The menu changes the displayed physical-code diagram. Compact hides navigation and numpad groups; ISO adds an extra physical position. Your operating system still determines the input language and character mapping.

Can I test my laptop or numeric keypad?

Yes, for key events that the browser reports to the focused area. Select a useful diagram and inspect code and location for numpad events. A laptop Fn layer or an OS-reserved key may never generate a page-visible event, so a missing result is not a hardware verdict.

Why can a dead key or input method look different here?

The tester records keydown meaning and code, not committed text or composition-event sequences. Browsers can report Dead or Process values, and canceling ordinary key defaults can affect composition. Use a normal text editor to test the final composed text.

What is the difference between Stop test and Reset highlights?

Stop test ends capture and clears current held state while retaining recorded rows, counts, maximum and last locks. Reset highlights stops and clears those accumulated results too. Copying or downloading stops capture and exports the retained result set.

Does resetting delete a report I already saved?

No. Reset clears the results held by this page. A downloaded JSON file or clipboard copy is separate and remains until you remove or replace it yourself. Review character values before sharing either copy.

Sources and references

Put it into practice.

Related guides