You can record your screen without installing anything by using the browser's own capture interface: open the page, press Record, and choose the screen, window or browser tab to share. A browser recorder cannot capture system or tab audio, so any narration has to come from your microphone.
The recording is produced as a WebM file with VP9 video where the browser supports it, written straight to your disk with no watermark, no account and no time limit beyond your machine's memory. Nothing is uploaded, so the footage stays on the device it was captured on. If you need MP4 instead, the result panel links to a converter that rewrites the file locally. The only thing to decide before you start is whether you want your voice on the recording: the microphone option is on by default, and you can switch it off for a silent capture.
The appeal of a browser recorder is that a support agent, a teacher or a colleague can send a link instead of a download. That is also where the disappointment starts, because the browser sandbox has real limits and most tool pages do not mention them. This guide covers what the recorder captures, what it cannot, and the cases where a desktop app is still the better answer.
Screen Recorder
Capture the screen, a window or a tab, add your microphone, pause when you need to, and download a WebM. No install and no watermark.
What a browser screen recorder can capture
A browser records the screen through one interface, getDisplayMedia, and that interface treats video and audio as two separate requests. This recorder asks for video and declines audio at that layer, then adds your microphone through a second, independent permission. The result is a capture with your voice if you want it and no sound from the machine itself.
The distinction matters because the sharing modes behave differently, and only some of that is obvious from the picker dialog.
| What you share | Captured | Notes |
|---|---|---|
| Entire screen | Video | Every window, notification and popup is in the file |
| A single window | Video | Resizing the window changes the captured area |
| A browser tab | Video | Still renders when the tab is behind another window |
| Microphone | Audio, optional | On by default, needs one permission prompt |
| System or tab audio | Not captured | The recorder requests video only at the capture layer |
Sharing a single tab is the mode most people should use for tutorials and demos. The browser renders that tab into the capture stream even when the window is hidden behind another one, so you can talk over a slide deck or a dashboard while your notes and your terminal stay on the physical screen. Sharing the whole screen does the opposite: every chat popup and every window switch ends up in the file, and a single notification can reveal something you did not intend to show.
How to record your screen without installing anything
- Press Record. The browser opens its own sharing dialog. Pick the whole screen, one window, or one tab. Leave the microphone box checked if you are narrating.
- Record, and pause when you need to. The timer runs in the corner of the preview. Pause stops both the capture and the clock, and resuming continues from the same point instead of starting a new file.
- Stop and download. The recording appears in the result panel with its size and duration. Download the WebM, or send it to the converter with one click if your destination wants MP4.
Three steps, no installer, no account and no trial watermark. The browser asks which surface to share on every single start and remembers nothing between sessions, which is the correct default for anything you might capture.
Choosing between screen, window and tab
Share a tab when what you are demonstrating lives inside one page: a dashboard, a code editor in the browser, a web app you are reviewing. Share a window when the demonstration spans one desktop application, such as a spreadsheet or a design tool. Share the entire screen only when the point is the desktop itself, for example showing how to install something or where a setting lives in the operating system. The narrower the surface, the less chance of publishing an accidental notification.
Recording with narration
The microphone is mixed into the same stream as the picture, so there is no separate audio track to sync later. Speak at a normal distance and level, since the recorder applies no gain control and no noise reduction: what your microphone hears is what ends up in the file. If the browser denies microphone access, the recording still proceeds with video only rather than failing, which is worth knowing the first time a demo comes out silent.
The limits, stated plainly
- No system audio and no tab audio. Sound playing on the machine is not captured at all. Only the microphone can be added, and only if the browser grants it.
- WebM out of the box, not MP4. The recorder writes WebM because that is what the browser's recording interface produces natively. Converting is a second step and a real re-encode.
- Nothing is editable here. There is no trim, no annotation, no webcam bubble and no cursor highlight. You get the capture as it happened.
- Length is bounded by memory. Frames accumulate in the tab until you press stop, so a very long session can slow the tab down or fail near the end.
- Desktop browsers only. Screen capture is not exposed to pages on Android or iOS, so a phone cannot use this workflow.
- The preview is muted. You see what is being captured, but you cannot monitor the microphone through the page without producing feedback.
None of these are bugs to be worked around. They are the shape of the platform, and knowing them in advance is the difference between a smooth recording and a second attempt.
When a desktop recorder is the better choice
Reach for an installed application, OBS being the common free one, in these three situations.
You need the sound of the machine
Recording a game, a video call, a system alert or a media player means capturing an audio device rather than a screen surface. A desktop recorder can do that because it runs outside the browser sandbox. A page cannot, and no amount of clicking will change it.
The session runs past half an hour
Web recordings live in tab memory, and an hour of footage is a large amount of it. Lecture capture, long interviews and full webinars belong in a tool that writes to disk as it goes. If a browser recording is the only option, break the session into ten minute takes and join them afterwards with the video merger, which concatenates clips without leaving the browser.
You need editing, a webcam overlay or a stream
Cutting dead time, drawing on the screen, adding a picture-in-picture camera or broadcasting live are all outside what a capture page does. In that case record locally, then edit the result. For the parts that are simple, the browser tools still help: the video trimmer cuts the ends off a take, and the video compressor brings a long recording down to a size a chat or an email will accept.
Recording a webcam instead of the screen
Some recordings are about a person rather than a screen: a standup update, a quick introduction, a message for a customer. That is a different capture source with the same privacy story, and the webcam recorder handles it in the browser with camera and microphone together. The rest of the workflow is identical, down to the WebM output and the download step.
How to confirm the file never leaves your device
The claim that a tool is private is easy to make and easy to check. Open your browser developer tools, switch to the network panel, clear the list and start recording. Watch the requests as you capture and stop: the page loads its own code, and the video frames never appear as an outgoing request, because the capture and the encoding both happen inside the tab. There is no upload step to wait for and no progress bar pretending to be one.
The same test on a converter site tells the opposite story: a request carrying tens of megabytes appears the moment you press convert, and you have no way to know when it is deleted. For a bug report that contains an internal tool, a customer's screen or part of a private chat, that difference is the whole point.