Yes, inside the code editor. As of September 9, 2026, HackerRank documents a Keystroke Codeplayer that replays how a candidate's code was typed, and an AI plagiarism model that scores keystroke patterns, code-writing patterns, solve time, and large deletions. What HackerRank does not document is a system-wide keylogger. In browser-based tests a web page only receives keys while it has focus, and the Desktop App Mode documentation describes editor-level typing analysis rather than logging every key pressed on the machine.
What HackerRank records from your typing
HackerRank's AI Plagiarism Detection documentation lists the signals its model analyzes: "Code writing patterns", "Time taken to solve the problem", "Copy-paste activity", and "Tab-switching behavior", along with keystroke patterns and code iterations. Two named patterns come directly from typing. "Suspicious Code Resetting" is described as large code deletions with irregular typing. "External Copy-Paste" is content brought in from outside the editor.
Reviewers see the result through a tool HackerRank calls the Keystroke Codeplayer, which replays the coding journey so a recruiter can watch the code appear in the order it was typed. The same documentation says flagged attempts are rated High or Medium, categorized by reason as Code Similarity or Coding Patterns, and that the model has "an overall precision of 85%, meaning flagged sessions are correct 85% of the time."
The Proctor Mode documentation adds one more typing signal: "Conversation detection in the code editor: Detects patterns where candidates type and delete content in the editor that may indicate external communication". Despite the name, this is not audio. It is a pattern in what you type and erase.
What a browser test cannot capture
A HackerRank test in Secure Mode or Proctor Mode runs as a web page. A page receives keyboard events only while it is the focused window. The browser does not deliver keys typed into another application to the test tab, so keystrokes in a separate editor, terminal, or chat window are not logged by the page.
They are not invisible, for a different reason. Default Tab Proctoring "Tracks how often and for how long a candidate leaves the test window", and the recruiter report shows "Out of Window Duration" and "Number of Window Exits" (Test Integrity documentation). In Proctor Mode, screenshots of the entire screen every 15 seconds capture whatever window you typed into. The keystrokes themselves are not recorded; the time away and the screen are.
What Desktop App Mode adds
HackerRank's Desktop App Mode documentation describes "operating system–level monitoring", but its typing claim is specific: suspicious typing patterns are "flagged for review as potential external assistance". The document does not describe capturing keystrokes outside the test interface. Since the app "Enforces full-screen mode throughout the test and does not allow candidates to navigate away from the test interface" and blocks other applications, there is little outside the editor to type into.
On macOS the app requests Accessibility and System Events permissions (candidate documentation). Accessibility is the permission macOS requires before an app can observe or control input across the system, so the app has the technical footing to see more than the editor. HackerRank does not document doing so. Treat that as an undocumented capability, not a confirmed one.
What a typing flag proves
Not much on its own. The model is a classifier with a published 85% precision, which still means roughly one flagged session in seven is wrong. HackerRank names the weak spot itself: the model struggles with "questions requiring minimal effort to achieve a full score or when solutions consist of very few lines of code."
Typing patterns that can trigger a flag without misconduct include pasting from your own scratch notes, deleting a failed approach and retyping a new one quickly, or typing a memorized solution at speed. The guide to what happens after an integrity flag covers how employers use these scores and how to respond if a recruiter asks you to explain your solution. The wider set of signals HackerRank combines with typing is laid out in how HackerRank detects cheating.
What this means for candidates
- Write in HackerRank's editor from the start rather than drafting elsewhere and pasting. A paste is a logged event; typing is a pattern.
- Expect the replay. If you can explain why you deleted and rewrote a block, the replay supports you.
- Read the consent line. HackerRank states "When AI plagiarism detection is enabled, candidates must consent before starting the test." If that line is on the setup screen, the model is on.
- If the rules prohibit outside help, use AI to practice before the test, not during it.
For the full inventory across modes, see what HackerRank proctoring detects and records and the companion note on copy and paste detection in HackerRank.
Share
Explore with AI
Continue exploring
Related guides
Assessment Guides · 5 min read
Can Codility Detect ChatGPT? Typing Patterns and AI Matches
Codility names ChatGPT and Claude in its Typing Pattern Detection and matches submissions against AI-generated solutions. What each layer flags and its limits.
Assessment Guides · 5 min read
Does HackerEarth Detect Cheating? What It Flags
HackerEarth flags code matching another candidate above 85%, runs AI checks against public sites, and scores an integrity index. What each flag proves.
Assessment Guides · 5 min read
Does CodeSignal Detect Cheating? Suspicion Score Explained
CodeSignal runs two checks: a human review of the proctoring recording and an automated Suspicion Score on editor activity. What each flags and what a flag proves.