WorldPreventer – Advanced Anti-Exploit & World Protection

⚠️ Early Development Notice WorldPreventer is still in a very early stage of development. Many systems are functional, but the machine learning (ML) detection system does not ship with a trained model.

• Do not enable ML detection unless you have trained your own model.
• If you see the notification REQUEST_STATS observed (WorldTools signature), this likely indicates that a client is attempting to download your world.

---

Overview

WorldPreventer is a high-performance anti-exploit and world protection plugin designed to safeguard your server from a wide range of threats, including:

• World downloads
• Chunk exploitation
• Suspicious client behavior

It combines packet-level monitoring, behavioral analysis, and adaptive responses to protect your server from malicious activity.

---

Features

Core Protection Systems

1. Chunk Exploit Prevention

   • Detects & prevents rapid chunk loading/unloading
   • Monitors chunk interaction patterns
   • Stops chunk boundary exploits

2. WorldTools Blocker

   • Blocks WorldTools client functionality
   • Prevents unauthorized world downloads
   • Detects & blocks rapid join/leave patterns

3. Deep Packet Inspection

   • Analyzes network traffic for suspicious patterns
   • Detects & blocks malicious packet sequences
   • Monitors data volume & timing anomalies

4. Adaptive Response System

   • Dynamic threat-level scoring
   • Escalating responses (warn → kick → ban)
   • Fully configurable actions

---

Machine Learning Detection (Optional)

Important: The ML system is disabled by default and ships without a trained model. Do not enable it unless you train your own.

When configured properly, the ML system provides:

• Anomaly Detection – Spotting unusual player behavior
• Behavioral Analysis – Learning normal server patterns
• Real-time Scoring – Constantly evaluating actions
• Adaptive Thresholds – Adjusting to server conditions

Tracked Features (12 total): Chunk transitions, container interactions, movement volume, entity spawn rates, packet bursts, systematic movement, join surges, client brand, timing precision, data volume variance, interaction patterns, movement entropy.

---

Commands & Permissions

Admin Commands

• /worldpreventer reload – Reload configuration
• /worldpreventer status – View protection status
• /worldpreventer simulate <player> <test> – Run simulation tests
• /worldpreventer ml [enable|disable|status] – Manage ML system

Permissions

• worldpreventer.admin – Full access & bypass
• worldpreventer.staff – Receive suspicious activity alerts
• worldpreventer.bypass – Exclude player from detection

---

Configuration

Example config.yml:

# Grace period after join
join_grace_period_seconds: 120

# Escalation actions
actions:
  on_exceed: WARN  # WARN, KICK, BAN

# Suspicion scoring
scoring:
  enabled: true
  decay_per_minute: 10.0
  thresholds:
    warn: 100.0
    kick: 500.0

# Machine Learning (disabled by default)
ml:
  enabled: false
  direct_kicks: false
  score_threshold: 3.0
  points_on_exceed: 20.0
  kick_message: "§cKicked by ML anti-cheat system"
  window_seconds: 60
  autosave_minutes: 5
  decay: 0.98

# ProtocolLib-based packet blocking
packet_blocking:
  enabled: true
  block_stats_request: false
  block_statistics: false
  block_map_data: false
  corrupt_block_entities: false
  inject_fake_data: false

---

Integration

Dependencies

• Required: ProtocolLib
• Recommended: Vault

API Example

WorldPreventer plugin = (WorldPreventer) Bukkit.getPluginManager().getPlugin("WorldPreventer");

// Get suspicion score
double score = plugin.getSuspicionScorer().getScore(player.getUniqueId());

// Add custom detection points
plugin.getSuspicionScorer().addScore(player.getUniqueId(), 10.0, "CUSTOM_EVENT");

---

Staff Notifications

Staff with permission receive alerts for:

• Suspicious activity
• ML system detections
• Config changes
• Warnings & errors

---

Bypass Options

Players excluded from detection:

• worldpreventer.bypass permission holders
• Players in Creative mode
• Players in Spectator mode

---

Support

For bug reports, feature requests, or support, please visit the GitHub repository.