Pepa

Advanced Behavioral Simulation System for WRobot
Overview
This combat class represents an experimental AI-driven Warrior Tank system designed for WRobot.
It combines predictive movement, adaptive hysteresis, and context-aware decision making to simulate highly realistic player behavior during combat.
The system is built around multiple synchronized threads controlling both rotation and movement intelligence independently, allowing it to react smoothly to complex combat situations and environmental changes.
Core Features
🧠 Adaptive AI Movement Engine
Dynamically positions the player at the threat centroid, repositions based on enemy spread, and adjusts movement smoothness according to FPS, ping, and encounter type (trash, elite, boss).
⚙️ Predictive Enemy Tracking
Utilizes Kalman-based filtering to estimate enemy movement vectors and accelerations, improving facing and threat control accuracy.
⚔️ Context-Aware Combat Positioning
Detects isolated targets, executes Charge or Intercept where appropriate, and uses fallback movement logic when gap closers are on cooldown.
🧩 Threaded Coordination System
Rotation and movement logic run in separate threads, synchronized through custom locks and adaptive state arbitration layers.
🧱 Adaptive Hysteresis System
Dynamically recalibrates movement thresholds based on performance metrics, preventing excessive micro-adjustments and improving realism under unstable frame rates or network latency.
🧭 Momentum-Based Movement Model
Simulates inertia and directional smoothing for more human-like motion transitions.
Entropy & Timing Source
This system uses a high-quality entropy source derived from the operating system scheduler and runtime timing variations.
Key properties:
The entropy is effectively CPU-free (it leverages natural scheduling jitter, timer granularities and other OS timing artifacts rather than expensive cryptographic operations).
The behavior produced by this entropy source is inherently non-deterministic and differs across machines — even identically configured systems will exhibit small, meaningful timing differences.
These timing differences are intentionally used to produce microvariation in movement and decision timing so that runs are not perfectly repeatable.
Note: the entropy mechanism is not a cryptographic RNG; it is a lightweight, practical source of behavioral randomness derived from normal OS operation and scheduling jitter.
Performance and Behavior
This class constantly self-adjusts based on:
Current FPS and ping
Number of enemies
Player health
Threat status and boss detection
It strives to behave differently under varying system conditions — introducing slight desynchronization across machines.
This creates unique microbehavior signatures, reducing the chance of pattern-based analysis producing identical traces.
⚠️ Important Notice — BETA Version
This system is currently in BETA and contains a carefully tuned internal race condition design.
These controlled timing variances serve as a behavioral anti-analysis layer, ensuring that no two executions behave identically — even on the same computer.
However, due to this design, the AI may sometimes:
Attempt to move in unintended directions
Overreact to sudden environmental or combat changes
Momentarily “break free” or exhibit erratic behavior (“self-destructive impulses”)
➡️ Manual supervision is absolutely required.
Never run this system unattended — it is intended for research and supervised use only.
Known Risks & Stability Notes
The system uses multiple threads and many fine-grained locks; while designed to be thread-safe, deadlocks or race outcomes can still occur under pathological conditions.
In extreme cases, the character may attempt risky movement (e.g., run directly into environmental hazards or pathing pitfalls).
Operator oversight is mandatory — monitor behavior and be ready to interrupt or take manual control.
 
🧩 Internal Architecture
Main (ICustomClass) │ ├── Thread #1 → RunRotation() │ - Handles combat, buffs, targeting │ ├── Thread #2 → RunMovementAI() │ - Handles positioning, facing, LoS, gap closer logic │ └── Sub-thread → PathfindingOptimizer - Calculates safe spots and escape routes     WarriorTankRotation │ ├── TankMovementAI │ ├── EnemyMovementPredictor ← Kalman filter prediction │ ├── AdaptiveHysteresis ← FPS / Ping adaptive smoothing │ ├── PathfindingOptimizer ← Separate thread for safe pathfinding │ ├── IsolatedTargetHandler ← Charge / Intercept decision system │ ├── ExponentialSmoothing ← Smooth motion and rotation │ ├── PerformanceMetrics ← Internal benchmark & debugging │ └── DangerState ← Environment threat evaluation │ ├── TurnCommandArbiter ← Rotational control priority │ └── WRobotCoordinator ← Manages shared movement ownership Combat Start │ ├─ Is manually controlled? → Pause AI │ ├─ Target in range? │ ├─ <5y → Maintain melee │ ├─ 8–25y → Use Charge/Intercept │ ├─ 5–8y → Fallback to charge range │ ├─ >25y → Direct approach │ ├─ Isolated from group? → Smart Gap Closer logic │ └─ Danger detected? ├─ Compute escape path (PathfindingOptimizer) └─ Defensive rotation & movement priority 🧾 Internal Evaluation
Category Score Architecture ⭐ 10/10 Movement AI ⭐ 9.5/10 Stability ⭐ 9/10 Combat Logic ⭐ 8.5/10 Human-like Control ⭐ 10/10 Total Score: 🧩 9.3 / 10
🧰 Requirements
WRobot for WoW 3.3.5a (WotLK)
.NET Framework 4.7.2+
Protection Warrior
💬 Final Words
This FightClass merges combat logic, movement prediction, and manual input detection into a smooth, human-like tank AI.
Perfect for soloing, grinding, or dungeons — safe, adaptive, and powerful.

🧪 [BETA] Advanced Blood Death Knight Tank 2.0 – Clean AI Movement & Smart Rotation
Author: Calaude
Class: Death Knight – Blood (Tank)
Version: 2.0 BETA (LoS + Debug Build)
Compatibility: WRobot for WoW 3.3.5a (Wrath of the Lich King)
🔥 Overview
This fight class is a fully re-engineered Blood DK Tank AI, designed to simulate realistic, intelligent combat behavior.
It features a complete custom movement AI, adaptive decision making, and multi-threaded control logic for smooth, human-like reactions.
Unlike traditional FightClasses, this one runs its own AI movement layer, coordinating with WRobot’s movement manager without conflicting or “fighting” for control.
🧠 Key Features
🧍‍♂️ Clean Movement AI
Fully autonomous positioning system adapted from the Warrior AI project
Predictive enemy tracking and movement anticipation using Kalman filtering
Intelligent facing, strafing, and backstep control
Adaptive hysteresis thresholds that adjust based on FPS and network latency
Smooth transitions between AI control and manual player input (WASD detection)
⚔️ Advanced Combat Behavior
Rotational logic for tanking: Icy Touch, Plague Strike, Death Strike, Heart Strike, Rune Strike, Blood Boil, Death and Decay
Full defensive cooldown usage: Icebound Fortitude, Vampiric Blood, Rune Tap, AMS, Bone Shield, Dancing Rune Weapon
Intelligent Death Grip logic that detects isolated targets and chooses the best pull strategy
Automatic Horn of Winter and buff upkeep
🧭 Smart Targeting & Positioning
AI constantly monitors enemy centroid, spread angle, and threat vectors
Repositions automatically when enemies surround the player or attack from behind
Pathfinding optimizer creates emergency escape paths based on enemy distribution
Handles corner cases like teleports, new pulls, and multi-target engagements
🧩 Defensive Awareness
“Danger Detection System” continuously evaluates:
Enemies casting or attacking from behind
Player health and incoming damage density
Safe zones for retreat or repositioning
Automatically engages defensive control when critical conditions are met
⚙️ Technical Highlights
Dual-thread architecture:
Rotation Thread – executes combat rotation
Movement Thread – runs AI movement logic
Thread-safe with proper locking and graceful shutdown
Built-in performance metrics for real-time profiling
Debug and diagnostic output for tuning behavior
🧰 Requirements
WRobot for WoW 3.3.5a
Death Knight level 80 (Blood Spec recommended)
Tanking role (defensive gear and presence)
English or localized client (spells auto-detected via string names)
🧪 Debug Options
Inside the code you can enable or disable specific debug sections:
    private const bool ENABLE_DANGER_LOGS = true; private const bool ENABLE_AI_DEBUG_LOGS = false; private const bool ENABLE_DEATH_GRIP_LOGS = true; Set these to true or false to toggle:
DANGER_LOGS: Prints warnings when surrounded, low HP, or in unsafe conditions
AI_DEBUG_LOGS: Detailed positioning and smoothing diagnostics
DEATH_GRIP_LOGS: Information about pull and engagement logic
🛠 Installation
Download the .cs file
Place it into:
    WRobot\FightClass\

Auto loot plugin  - Auto loot near enemy corpses

Warrior TANK autokick

The Deviate Fishes in the Barens