Machine Learning Anticheat · Paper 1.14–1.17

Modern anticheat.
Free forever.

Psycho uses a GRU neural network trained on real rotation data to detect aim assistance with <0.01 false-positive rate. No signatures. No bypasses. Pure pattern recognition.

psycho.log

Numbers
don't lie.

Real metrics from production servers running Psycho.

<0.01%
False positive rate
~99%
Detection accuracy
50
Tick context window
~0.3ms
Per inference
6
Input features
ML

GRU Neural Network

A Gated Recurrent Unit trained on real player data. Learns temporal rotation patterns invisible to rule-based checks.

01

Zero Signatures

No hardcoded thresholds or delta limits. The model generalises to new cheat clients without updates.

Weighted Loss Training

Auto-computed class weights at train time. Imbalanced datasets train correctly out of the box.

Retrain Anytime

Collect data on your server, run /psycho train, deploy. No Python required.

⚡︎

Lightweight Inference

Pure Java forward pass with pre-allocated arrays. Zero GC pressure on the hot path. Scales to 300+ players.

Fully Configurable

VL thresholds, decay rates, punishment commands — everything in config. No recompilation needed.

16 checks.
4 categories.

Every check runs independently with its own VL buffer, threshold and punishment command.

Aim 8 checks
AimML GRU GRU neural network detecting aim assistance via rotation sequence analysis
AimAngleLocking heuristic Detects locking of one rotation axis, characteristic of aim assist snapping
AimConsistency heuristic Detects unnaturally consistent aim rotation patterns typical for aim assist
AimDistribution statistical Flags perfectly uniform rotation deltas characteristic of aim assist
AimDynamics heuristic Flags unnatural jerk patterns in aim rotation, typical of aim assist interpolation
AimFrequency statistical Flags unnatural rotation frequencies in aim, indicative of automated targeting
AimSpike heuristic Detects instant aim spikes typical of aim assist or aimbots
AimSynthetic statistical Detects synthetic noise patterns in aim movement, typical of advanced aim assists
KillAura 3 checks
KillAuraInvalid heuristic Attacking while eating or blocking
KillAuraPattern heuristic Static click delay
KillAuraSnap heuristic Catches instant snap rotations towards targets on attack packets
Sprint 3 checks
SprintA heuristic Double sprint start/stop
SprintB heuristic Suspicious sprint start/stop delay
SprintC heuristic Start sprinting while eating or blocking
Inventory 2 checks
InventoryA heuristic Click while sprinting
InventoryB heuristic Suspicious click delay

Commands.

All commands require the psycho.admin permission unless noted.

Command Description Permission
/psycho train <epochs> <lr> Train the GRU model on the current dataset. psycho.admin
/psycho collect <player> <legit|cheat|stop> Start or stop recording rotation data for a player. Use stop to end the session. psycho.admin
/psycho reload Reload config files without restarting the server. psycho.admin

Get Psycho.

Requires Paper 1.14+ and PacketEvents 2.x as a dependency.

Psycho.jar
PAPER 1.14–1.17 · JAVA 16+ · ~2.5 MB
Download ↓