How PhantomCodeAI stays invisible
The undetectability claim is enforced at the operating-system level, not by the application's UI. When PhantomCodeAI creates its overlay window, it sets a capture-exclusion flag understood by every modern video-conferencing app, screen recorder, and remote-proctoring agent on the platform.
- macOS:
NSWindowSharingNonetells Core Graphics to exclude the window from any screen-capture API call (CGDisplayStream,SCStream, ScreenCaptureKit). Zoom, Google Meet, Teams, OBS, QuickTime, Loom, and Apple's own screen-recording shortcut all use these APIs. - Windows:
WDA_EXCLUDEFROMCAPTUREon the window's display affinity, set viaSetWindowDisplayAffinity. Same effect on every conferencing and recording tool that uses the standard Windows capture pipeline (BitBlt, DXGI Desktop Duplication, Windows Graphics Capture).
The overlay also draws with click-through transparency (NSWindow.ignoresMouseEvents on macOS, WS_EX_TRANSPARENT on Windows), so cursor input goes through to the interview application underneath. The interviewer's mouse never “catches” on the overlay even if it somehow became visible.
What “undetectable” does not mean
No software is undetectable by every conceivable inspection method. PhantomCodeAI is undetectable by the tools an interviewer or proctor actually uses during a live interview — screen sharing, screen recording, proctoring agents, and the visible process list. It is not designed to evade root-level forensic inspection by someone with full local administrative control of your machine, and that situation does not arise in a remote technical interview.
Verifying it yourself
The fastest way to confirm undetectability is to start a Zoom call with a friend (or join an empty Zoom meeting and record it locally), share your screen, and toggle PhantomCodeAI on and off. The window appears on your physical display but is absent from the recording. PhantomCodeAI also ships with a built-in “test undetectability” mode that walks you through this check during onboarding.