How to Show Keyboard Shortcuts On Screen in Your Recordings

Keyboard input is invisible on screen. Here is how Recorded captures the keys you press during a recording and turns them into editable on-screen overlays.

You record a clean two-minute walkthrough. A panel opens, a selection expands, a file saves — and nowhere on screen is there any hint that you pressed Command-Shift-P, Shift-Down, or Command-S to make those things happen. To you it was a sequence of deliberate keystrokes. To the viewer, the UI moved by itself.

This is the single most common gap in technical screen recordings, and it is structural: the screen only shows results, never input. Recorded closes it by recording the keys you press alongside the video, then turning them into text overlays you can edit. Here is exactly which keys it captures, what the overlays look like, and how to clean them up before you export.

Why Invisible Input Breaks a Tutorial

A viewer following along has to reproduce what you did. When the input is a click, they can see the target. When the input is a keystroke, they have three bad options: guess, pause and search the docs, or give up and watch passively.

The cost shows up in specific places:

  • Command palettes and launchers. The palette appearing is the visible part. The shortcut that opened it is the part worth copying.
  • Modifier-click behavior. A Shift-click that extends a selection and a plain click that replaces it look identical on screen.
  • Terminal work. Control-C, Control-R, and tab completion all produce output with no visible cause. The terminal and CLI demo guide covers the rest of that problem.
  • Form and editor navigation. Tab order, arrow-key navigation, and Escape-to-dismiss are invisible by definition.

Narrating the keys helps, but narration disappears the moment someone watches with the sound off — which, on most social and in-app players, is the default. Overlays survive that. See designing screen recordings that work with the sound off for the broader version of this argument.

What Recorded Captures While You Record

Keystroke capture runs automatically for the whole length of every recording. There is no toggle to flip before you start, and it does not depend on which capture mode you chose.

Two categories of keys are recorded:

Keys that are always captured. F1 through F12, Escape, Tab, Enter, Backspace, Space, Caps Lock, Num Lock, Insert, Delete, Home, End, Page Up, Page Down, and the four arrow keys. These are the keys that drive software without producing visible text, so they are worth showing on their own.

Letters, numbers, and symbols — only while a modifier is held. A, 5, or / pressed by itself is not recorded. The same key pressed with Control, Option/Alt, or Command/Win is. So Command-C is captured; typing the word “command” is not.

That distinction is the privacy design, not a limitation. Typing a password, a customer’s email address, or a draft message into a field produces no keystroke record at all, because none of those involve a non-Shift modifier. It is a meaningful safeguard, though it is not a substitute for the rest of a careful workflow — the privacy and security guide covers what else to close and clear before you hit record.

Modifier keys themselves are also tracked, so a standalone Option press that reveals an alternate menu shows up even though no other key followed it.

macOS Needs Accessibility Permission; Windows Does Not

On macOS, reading key events system-wide requires Accessibility permission, so Recorded asks for it on first launch alongside Screen Recording permission. The permission screen lists both, opens the right System Settings pane for each, and re-checks every couple of seconds — once both are granted it moves on to the main window by itself, with no app restart.

If Accessibility was never granted, the recording still works fine; it simply contains no keyboard data, and no key overlays appear in the editor. That is the usual explanation for “the keys didn’t show up” on a Mac.

On Windows, keyboard state is polled directly and no special permission is involved. Capture works as soon as you start recording.

What the Overlays Look Like in the Editor

Open a new recording in the editor and the keystrokes are already there, converted into text segments on the text track of the timeline. You are not building them by hand; you are editing something that already exists.

The defaults are chosen to be readable without tuning:

  • Position: bottom-center, clear of most UI chrome
  • Duration: 1.5 seconds per keystroke, trimmed early if the next keystroke arrives sooner
  • Style: bold white text on a black background at 80% opacity, with padding and rounded corners
  • Size: 20px at 1080p, scaled up automatically for higher-resolution recordings
  • Animation: a 0.15-second fade in and out

Keys pressed within 0.3 seconds of each other are merged into one overlay, so a fast Command-C followed by Command-V reads as a single group instead of two flashing labels.

Modifier names are rendered the way each platform writes them. A recording made on macOS shows ⌘ + ⇧ + P; the same shortcut recorded on Windows shows Ctrl + Shift + P. Arrow keys become ↑ ↓ ← →, and long names are shortened — Escape to Esc, PageDown to PgDn, Delete to Del.

Cleaning Up the Pass That Matters

Automatic capture is thorough, which means it captures keys you do not want to feature. A single cleanup pass is what separates a useful overlay track from visual noise.

Delete the typing debris first. Space, Backspace, and Enter are always captured, so any stretch where you typed prose into a field leaves a trail of Space and Backspace labels. Select them on the text track and press Delete or Backspace, or use the right-click menu. Multi-select works, so a long run goes in one action.

Keep only the keystrokes that teach something. A good rule: keep it if a viewer would need to reproduce it, or if the on-screen result is otherwise unexplained. Drop it if it was navigation you happened to do with the keyboard — scrolling a page with Page Down rarely needs a label.

Fix the ones that fire too fast. When you trigger two shortcuts in quick succession, the first overlay is truncated at the moment the second starts. If both matter, drag the segment edges on the timeline to give the first one more room, or space the actions out in the next take.

Move overlays that collide with content. Bottom-center sits on top of status bars, terminal prompts, and captions. Select the segment, open the Text tab, and pick one of the nine positions — top-center and top-right are the usual escapes.

Restyle once, apply to the rest. The Text tab controls content, font size, font family, weight, alignment, text and background color, padding, corner radius, and the transition. If you are building a series, settle on one look and reuse it so episodes match, which is the point of a team video style guide.

You can also edit the text itself. Changing ⌘ + S to ⌘ + S — save costs nothing and removes the last bit of ambiguity for a viewer who does not recognize the symbols.

Recording Habits That Make Key Overlays Land

The overlay shows up at the instant you pressed the key. Whether the viewer connects it to the result depends on how you recorded.

Pause before the shortcut, not after. Stop moving, let the screen settle, then press the keys. If the cursor is mid-flight when the overlay appears, attention splits.

Hold still for a beat afterward. The overlay needs its 1.5 seconds and the UI needs time to respond. Clicking immediately afterward buries both.

Do not chain shortcuts you intend to explain. Three shortcuts in a second produce one merged label and no comprehension. Perform them as three separate, deliberate actions.

Say the key name out loud anyway. Overlays serve sound-off viewers; narration serves everyone else. They cost nothing together.

Zoom on the result, not the overlay. Zoom regions apply to the captured frame while text overlays are drawn on top, so a tight zoom on the affected panel does not push the key label off screen.

When to Turn Them Off Entirely

Key overlays are not right for every video. Delete the whole text track when:

  • The recording is a marketing or launch clip, where a technical label stream undercuts the polish
  • You are showing outcomes rather than methods — an executive update does not need the shortcut you used to open the dashboard
  • The video is going out as a GIF, where the frame rate is low and small text is the first thing to break down
  • Another overlay already owns the lower third, such as burned-in captions

Deleting them is non-destructive. The keyboard data stays with the recording, so you can rebuild the overlays in a future edit if you change your mind about how the video should read.

A Short Checklist

Before export, run through this:

  1. Watch once at full size with the sound off and ask whether every UI change has a visible cause
  2. Delete Space, Backspace, and Enter overlays left over from typing
  3. Confirm no overlay sits on top of a status bar, prompt, or caption
  4. Extend any shortcut that got truncated by the keystroke after it
  5. Check that modifier symbols match the platform your audience uses
  6. Spell out anything ambiguous in the overlay text itself

For the broader craft of on-screen text — callouts, labels, and timing — see using text overlays and annotations. If you want to try a keystroke-heavy take on your own machine, download Recorded for macOS or Windows, record thirty seconds of your favorite shortcut sequence, and watch how much of the work suddenly becomes visible.