Record your screen, camera or microphone to a file REEL

A screen recording is a copy of your screen

That sentence sounds obvious and is worth saying, because the usual way to record a screen is a service that captures it and hosts the result. What you have then is a video of your desktop, with whatever was in the other windows, the notification that arrived mid-recording, the tab titles, the client name in the corner, sitting on somebody else's server behind a link that anyone holding it can open.

People record screens for support tickets, bug reports, handovers and walkthroughs, which means the screen usually has real work on it. And the recording is often made by the person least able to say whether it is allowed: somebody trying to show a colleague a problem, at speed, because explaining it in words was not working.

Here the browser's own MediaRecorder does the capture. The stream is encoded as it arrives (WebM with VP9 or VP8 and Opus, or MP4 in Safari) the chunks are held in memory until you stop, and then handed to you as one file. No server sees a frame, because there is no request that would carry one.

Before you press record

Close what you are not showing, and turn off notifications properly rather than hoping. A recording is the one place a message preview appears with someone's name and the first line of what they said, and it is impossible to take back once the file has been sent.

Choose the narrowest capture that does the job. A single window shows one window; the whole screen shows everything, including whatever you switch to by reflex. A browser tab is narrower still. The dialogue asks you to choose, and the default is not the safe one.

Decide about the microphone deliberately. Screen and voice is what makes a walkthrough useful, and it also records the room: a colleague's remark, a phone call, a child. If in doubt record the screen silently and add a written note, which is easier to edit afterwards than audio.

Then watch it back before you send it. The whole recording, not the first ten seconds. It costs a minute and it is the only step that catches what you did not notice at the time.

What it does not do

It does not edit. If you need to cut the first thirty seconds or the bit where you found the right menu, this is not the tool. SNIP trims audio, and video editing is a different program. The practical answer is to plan the take: know what you are going to show, then record it once, which is usually faster than editing anyway.

And it cannot capture a tab's own audio on a browser that does not offer it, which varies by browser and by what you chose to share. The page says which it got, rather than recording silence and letting you find out afterwards. Check the indicator before the take that matters.

For the file afterwards: AVSCRUB strips its metadata, RECAST blacks out a region of the picture permanently, HUSH bleeps words out of the audio, and SCRIBE transcribes it on the device if what you actually wanted was the words.