Can HackerRank see your screen?

You are about to open a timed assessment and you want to know whether that tab can watch the rest of your machine. This section draws the actual line. The boundary is the browser itself, and the only way past it is a permission dialog you have to click.

A web page runs inside the browser sandbox. It can see what happens inside its own tab, and the browser tells it when that tab loses focus or leaves full screen. It cannot read the pixels of your desktop, list the applications you have open, or reach a second monitor. The one interface that captures a screen, getDisplayMedia, always raises a system dialog where you choose a tab, a window, or a whole screen, and nothing is captured until you pick one.

So the question has an answer you can verify yourself, without trusting anyone's marketing: if no permission dialog appeared, no screen capture is running. Related pages on what interviewers can and cannot observe are grouped under detectability.

Proctoring on HackerRank is optional and configured per test by the hiring team, so two assessments from the same company can behave differently. When it is switched on, the documented signals are webcam snapshots, tab and window focus changes, full-screen exits, copy-paste events, and code-similarity scoring across submissions.

Read that list again. Every item is either a browser event or the code you submitted. Focus logging tells a reviewer that you left the tab and for how long. It does not tell them where you went, which application you opened, or what sat on the other monitor. The distinction matters, because a focus log reads as alarming in a report while carrying far less information than most candidates assume.

• Focus and full-screen events: that you switched away, not what you switched to.
• Copy-paste: that text arrived by paste, not where the text came from.
• Webcam snapshots: only when the employer enabled the camera, and only after your browser asked for it.
• Plagiarism scoring: similarity between your submission and other submissions, read by a person.

There are real cases where the answer flips to yes. HackerRank CodePair is a live session with a human on the other side, and an interviewer can ask you to share your screen. A recruiter may also run that conversation over Zoom, Google Meet, or Microsoft Teams next to the editor. Once you choose a surface in the share dialog, everything on that surface is visible and often recorded.

A company-managed laptop is the other case. Device management software runs outside the browser and none of the boundaries above apply to it. If the machine belongs to an employer, assume the employer can see it.

SubcueAI states this plainly rather than around it: screen share, screen recording, proctored setups, and company-managed devices sit outside what any assistant can help with. How audio and transcripts are handled is written up on the security page.

SubcueAI ships two live-assist surfaces, and neither one injects anything into the HackerRank page. The desktop app for macOS and Windows captures system audio and your microphone and draws a floating overlay on your own machine. The browser extension works from a Side Panel and captures the audio of a meeting tab, the interviewer's side only. No bot joins the call.

Consider a backend engineer interviewing at a payments company: a timed HackerRank assessment first, then a CodePair round with an engineer. In the timed assessment nobody is speaking, so a live audio assistant has nothing to work from, and that round is won by preparation instead. In the CodePair round, if they are asked to share their screen, the honest answer is that a screen share shows whatever they shared.

Which puts the useful work before either round. Timed practice against a mock interviewer rehearses the part that decides both, explaining your reasoning out loud while you write code. Setup for the live rounds is walked through on the tutorial page.