Single-file, offline HTML tool by Vektor T13 Technologies for detect.expert.
Developed by Dmytro Momot.
This page captures mouse telemetry in a sandboxed field and performs real-time anomaly detection based on movement kinematics and behavioral patterns. It can be used as a lightweight behavioral/browser fingerprint signal (mouse input profile), especially when combined with Pointer Lock (relative-only deltas).
-
Mouse telemetry capture
- local/client/screen coordinates
movementX/movementYdeltas (relative input)isTrusteddetection (flags scripted/injected events)
-
Real-time kinematics
- Δt, Δr, speed (px/s)
- |acceleration| and |jerk|
- mini speed chart (last ~2 seconds)
- path visualization + teleport markers
-
Anomaly heuristics
- “teleport” / unrealistically fast displacement
- extremely smooth high-speed segments (too clean relative stream)
- high linearity (R²) or abnormal angle entropy
- mismatch between
movementX/Yand Δposition (absolute vs relative profile)
-
Live analysis panel
- R² path linearity score + interpretation
- direction entropy (Shannon) + interpretation
- absolute vs relative quality metric
- last events table (max 50) with flags
-
i18n language switcher (top-right)
- Русский / English / 中文(简体) / Tiếng Việt
- Language preference is stored locally via
localStorage.
-
Offline / privacy-friendly
- No external dependencies.
- Network is blocked via CSP (
connect-src 'none').
- Open the HTML file in a modern browser (Chromium / Firefox / Safari).
- Move the mouse inside the tracking field.
- Click the field to enable Pointer Lock (press Esc to exit).
- Watch metrics/flags update in real time.
- Switch language in the top-right selector.
- Use Reset to clear in-memory data.
- Pointer Lock provides cleaner relative deltas (
movementX/Y) and makes mismatch analysis more informative. isTrusted = falsetypically indicates programmatic event injection (synthetic input).- Thresholds are heuristic/empirical and intended for practical detection/triage, not as a strict proof.
© Vektor T13 Technologies • detect.expert • Developed by Dmytro Momot