公告:本项目将不再更新
AtomGuard
让您的 Minecraft 服务器免受黑客、机器人和崩溃的侵害。
兼容 Paper 1.21.4 和 Velocity 3.x
AtomGuard 是一款免费且开源的安全插件,可保护您的 Minecraft 服务器免受攻击、漏洞利用和作弊者的侵害——让您能专注于构建社区。
⬇️ 下载 · ? 设置指南 · ❓ 常见问题 · ? 贡献
? 为什么选择 AtomGuard?
运行 Minecraft 服务器很有趣——直到有人试图破坏它为止。向服务器涌入虚假玩家的机器人攻击、导致所有人掉线的漏洞利用、摧毁您经济系统的复制故障……此类问题不胜枚举。
AtomGuard 为您处理所有这些麻烦。只需将其放入 plugins 文件夹,您的服务器便受到保护。无需复杂的设置,也不需要计算机网络学位。
它能为您提供以下防护:
?️ 机器人与 DDoS 攻击 — 自动检测并阻止大量虚假玩家和连接洪水,防止它们拖慢您的服务器速度。
? VPN 和代理滥用 — 阻止被封禁的玩家通过 VPN 重新加入。使用 7 种不同的检测方法以确保准确性,并且不会误封您的正常玩家。
? 崩溃漏洞利用 — 修复了 44 种以上已知的 Minecraft 漏洞利用(黑客常用来使服务器崩溃),包括书籍崩溃、NBT 攻击和数据包漏洞利用。
♊ 复制故障 — 防止可能毁坏服务器经济系统的物品复制错误。
⚡ 卡顿机器 — 限制红石、爆炸、活塞和实体的数量,防止玩家建造导致服务器卡顿的设备。
? 国家筛选 — 如果您希望建立地区锁定的社区,可以允许或阻止来自特定国家的玩家。
? 登录保护 — 阻止暴力破解密码攻击,并禁止使用常见密码(可与 AuthMe 配合使用)。
✨ 功能亮点
? 智能机器人检测
AtomGuard 不仅仅依赖单一指标——它会针对每个连接的玩家根据 8 种不同的信号进行评分,例如连接速度、客户端类型、加入模式和用户名。得分较低的正常玩家会立即通过。可疑玩家会被要求进行简单的数学挑战。明显的机器人则会被直接拦截。
您的普通玩家甚至不会察觉它的存在。
⚔️ 自适应 DDoS 防护
SmartThrottle 引擎会根据攻击的严重程度自动调整防御措施:
? 正常 → ? 升高 → ? 高 → ? 严重 → ⛔ 封锁
在遭受严重攻击时,它会锁定新连接,同时确保经过验证的玩家仍能进入。一旦攻击停止,一切会自动恢复正常。
? 威胁情报
AtomGuard 会在 7 天内学习您服务器的正常流量模式。当发生异常情况时——例如凌晨 3 点连接数突然激增——它会自动加强防护。这意味着更少的误报和对真实威胁更快的响应。
? 玩家信任度评分
玩家在您的服务器上正常游玩且不惹事的时间越长,AtomGuard 对他们的信任度就越高。资深玩家将跳过大部分检查,这意味着忠诚的社区成员登录更快,而新玩家将面临更严格的检查。
| 等级 |
对象 |
效果 |
| ? 新手 |
刚加入 |
执行完整安全检查 |
| ? 常客 |
已游玩一段时间 |
标准检查 |
| ✅ 可信 |
记录良好 |
跳过攻击模式下的检查 |
| ⭐ 资深 |
长期玩家 |
也跳过机器人和 VPN 检查 |
? 蜜罐
AtomGuard 可以在另一个端口上运行一个假的 Minecraft 服务器。当机器人扫描器发现它并尝试连接时,它们的 IP 地址会被立即列入您真实服务器的黑名单。这就像是为攻击者设置的陷阱。
? 攻击报告
每次攻击结束后,AtomGuard 都会保存一份详细报告——发生了什么、何时发生、屏蔽了多少个 IP 以及哪些防御措施发挥了作用。您可以随时查看这些报告,了解您的服务器曾面临过什么。
? 所需环境
| 项目 |
版本 |
是否必需 |
| ☕ Java |
21 或更高 |
✅ 是 |
| ? Paper 服务器(或其分支如 Purpur) |
1.21.4 |
✅ 是 |
| ? PacketEvents 插件 |
2.6.0+ |
✅ 是(核心插件必需) |
| ? Velocity 代理 |
3.x |
仅在使用代理时必需 |
| ?️ MySQL 数据库 |
8.0+ |
可选 — 用于持久化数据 |
| ? Redis |
7.x |
可选 — 用于多服务器同步 |
| ? MaxMind 许可证密钥 |
— |
仅在使用国家筛选功能时必需 |
? 快速入门
选项 1:Paper 服务器(大多数人从这里开始)
步骤 1 — 下载 PacketEvents 并将其放入 plugins/ 文件夹。
步骤 2 — 下载 AtomGuard-core.jar 并将其放入同一个 plugins/ 文件夹。
步骤 3 — 启动(或重启)您的服务器。AtomGuard 将自动生成其配置文件。
步骤 4 — 如果需要,自定义 plugins/AtomGuard/config.yml 中的设置——但默认值开箱即用,效果极佳!
选项 2:Velocity 代理
如果您运行的是 Velocity 网络,请安装代理模块以实现全网保护:
第 1 步 — 下载 AtomGuard-velocity.jar 并将其放入代理服务器的 plugins/ 文件夹中。
第 2 步 — 启动(或重启)您的代理服务器。配置文件将自动生成。
第 3 步 — 根据需要编辑 plugins/atomguard-velocity/config.yml。
第 4 步 (可选) — 如果您希望代理服务器和后端服务器共享封禁列表及威胁数据,请在两个配置文件中启用 Redis 部分。
? 提示: 您可以同时运行这两个模块!Velocity 模块在网络边缘阻止威胁,而 Core 模块则处理每个服务器上的游戏内漏洞利用。
? 命令
除非另有说明,所有命令都需要 atomguard.admin 权限。
| 命令 |
功能 |
/atomguard status |
查看正在运行的模块及其统计数据 |
/atomguard reload |
重新加载配置而无需重启 (需要 atomguard.reload) |
/atomguard stats |
查看统计信息仪表板 |
/ag intel status |
检查当前威胁等级 |
/ag trust info <player> |
查询玩家的信任分数 |
/ag replay list |
浏览过去的攻击报告 |
/ag honeypot status |
检查蜜罐是否处于活动状态 |
/panic |
? 紧急锁定 — 阻止所有新连接 (需要 atomguard.panic) |
权限
| 权限 |
功能 |
atomguard.admin |
完全访问所有命令 |
atomguard.bypass |
玩家跳过所有安全检查(请授予此权限给工作人员) |
atomguard.notify |
当漏洞被阻止时,玩家会收到聊天警报 |
atomguard.reload |
可以重新加载配置(除此之外无其他权限) |
atomguard.panic |
可以触发紧急锁定 |
? VPN 检测工作原理
AtomGuard 不依赖单一提供商(这可能导致不准确),而是同时查询 7 个提供商,并且只有当至少 2 个提供商确认该 IP 为 VPN 或代理时才进行封禁。这极大地减少了误报——使用正常互联网连接的玩家不会被封禁。
查看 7 个提供商
| # |
提供商 |
检测方法 |
| 1 |
本地黑名单 |
您自己的自定义 IP 列表 |
| 2 |
CIDR 拦截器 |
IP 段规则 |
| 3 |
DNSBL |
Spamhaus、DroneBL 以及自定义 DNS 黑名单 |
| 4 |
IPHub |
商业 VPN/代理数据库 |
| 5 |
ProxyCheck.io |
实时代理检测 API |
| 6 |
AbuseIPDB |
检查 IP 滥用历史 |
| 7 |
IPApi |
检查 IP 是否属于托管/数据中心提供商 |
干净的 IP 会被缓存,以便返回的玩家不再次进行检查。
? 技术细节
这些部分面向想要确切了解幕后运作机制的服务器管理员。
?️ Velocity 代理模块 — 详细信息
Velocity 模块在连接到达后端服务器之前,就在代理层面拦截连接。
连接保护
- SYN 洪水检测器 — 阻止每秒超过连接阈值的 IP
- Slowloris 检测器 — 识别慢速滴漏式连接耗尽攻击
- 流量异常检测器 — Z-score、缓慢爬升和脉冲攻击检测
- 连接指纹识别器 — 协议 + 主机名 + 时间戳指纹,用于检测机器人军团
- 子网分析器 — 在 /24 和 /16 级别检测协调一致的僵尸网络
- 滑动窗口速率限制 — 在每 IP、每子网和全局级别强制执行
- 验证玩家护盾 — 在严重/锁定期间为干净玩家保留一个槽位
- 攻击会话记录器 — 完整的会话日志,支持 JSON 导出
机器人评分细分
| 信号 |
权重 |
| 连接速度 |
20% |
| 加入模式 |
20% |
| 握手有效性 |
15% |
| 客户端品牌 |
15% |
| 地理/国家 |
10% |
| 用户名模式 |
10% |
| 协议版本 |
10% |
分数阈值: < 40 = 通过,40–60 = 标记,60–75 = CAPTCHA 挑战,75–90 = 踢出,90+ = 自动封禁。
额外层级: 品牌分析器(白名单包含 Fabric、Forge、Lunar、Badlion、LabyMod、OptiFine、Sodium)、昵称拦截器(正则表达式模式、长度限制、特殊字符分析)、验证玩家缓存(为干净玩家提供 48 小时豁免)。
聊天与命令保护
聊天速率限制、重复消息检测、Tab 补全洪水阻止、命令垃圾邮件预防、切换服务器滥用预防、数据包大小限制、崩溃循环检测(30 秒内断开连接 3 次以上)以及短会话标记(< 3 秒)。
账户保护
IP 声誉评分(随时间衰减,奖励干净登录)、带宽限期的自动封禁引擎、临时封禁管理器、Mojang 账户验证以及可热重载的 JSON 黑名单/白名单。
内核级阻止 (IPTables)
可以直接将封禁推送到 iptables、ip6tables 或 nftables — 被阻止的流量甚至不会到达 JVM。支持 /24 级别的子网封禁。启动和关闭时自动清理。
国家/地理过滤
MaxMind GeoIP2 集成 — 白名单或黑名单整个国家。每周自动更新数据库。
密码安全 (AuthMe)
5 次登录失败后临时封禁,阻止 10,000+ 种已知弱密码,检测同一 IP 下的密码相似性。
? Core 插件 — 详细信息
Core 插件在每个 Paper 服务器上运行,并通过 44+ 个模块处理游戏内安全。
数据包与网络漏洞利用
在 Netty 管道级别进行无效数据包过滤、阻止过大的数据包、防止离线数据包注入、检测数据包计时与延迟滥用。
NBT 与物品攻击
嵌套 NBT 深度限制、过大的 NBT 负载检测、Bundle 崩溃预防、对所有库存操作进行物品清理。
世界与区块崩溃者
修复书籍与讲台漏洞、修复地图标签崩溃、修复物品展示框崩溃、防止告示牌漏洞、区块崩溃保护。
修复复制漏洞
背包复制、物品栏点击复制、牛与骡子复制,以及通用的防复制引擎。
性能限制器
红石电路速率限制、爆炸限制、活塞限制、掉落方块限制、每区块实体限制。
AtomShield™ 行为分析
针对每位玩家分析 9 项信号:连接速率、重力验证、数据包时序、延迟与握手、协议、用户名模式、首次加入行为、加入后行为以及启发式画像。构建实时行为档案。当 TPS 下降或检测到连接洪水时,自动激活攻击模式。
威胁情报引擎
168 小时 EMA 流量画像(7 天滚动基线),跨 3 个威胁等级的 Z-score 异常检测,升级前 3 分钟确认窗口(防止误报),关键异常时自动进入攻击模式。命令:/ag intel <status|reset>
玩家信任评分
EMA 加权公式:游戏时长 + 干净会话 + 违规历史。通过 trust-scores.json 进行持久化存储。命令:/ag trust <info|set|reset|top>
取证分析
包含 UUID、时间线、峰值速率、被屏蔽 IP 和按模块统计的攻击快照。4 个严重程度等级(LOW/MEDIUM/HIGH/CRITICAL)。自动导出至 forensics/attack-<uuid>.json。命令:/ag replay <list|latest|<id>|export>
蜜罐模块
模拟的 TCP Minecraft 服务器(SLP 协议),用于诱捕机器人扫描器。自动将探测 IP 列入黑名单。命令:/ag honeypot <status|stats>
⚡ 集成
| 集成 |
功能说明 |
| MySQL + HikariCP |
使用连接池实现持久化数据库存储 |
| Redis Pub/Sub |
在整个网络中同步封禁、威胁数据和信任评分 |
| Discord Webhooks |
发生攻击时获取即时 Discord 通知 |
| Web Panel |
基于浏览器的仪表板,提供实时统计数据 |
| 配置迁移 |
更新时自动升级配置文件——并创建备份 |
| 异步日志记录 |
所有日志记录均在主线程之外进行,支持 7 天轮转 |
| 热重载 |
更改设置无需重启服务器 |
? 开发者 API
想要集成 AtomGuard 或在其基础上构建?我们提供了完整的 API。
查看 API 文档
Maven 依赖
<dependency>
<groupId>com.atomguard</groupId>
<artifactId>AtomGuard-api</artifactId>
<version>1.2.2</version>
<scope>provided</scope>
</dependency>
快速示例
// 检查 IP 的信誉评分
IReputationService rep = AtomGuardAPI.getInstance().getReputationService();
int score = rep.getScore(player.getAddress().getAddress());
// 在运行时启用或禁用某个模块
IModuleManager modules = AtomGuardAPI.getInstance().getModuleManager();
modules.setEnabled("bot-koruma", false);
// 访问信任评分、取证和情报数据
AtomGuardAPI.getInstance().getTrustScoreManager();
AtomGuardAPI.getInstance().getForensicsManager();
AtomGuardAPI.getInstance().getIntelligenceEngine();
// 监听事件
@EventHandler
public void onExploitBlocked(ExploitBlockedEvent event) {
String module = event.getModuleName();
Player player = event.getPlayer();
}
可用事件
| 事件 |
触发时机 |
ExploitBlockedEvent |
漏洞利用被拦截时 |
AttackModeToggleEvent |
攻击模式开启或关闭时 |
PlayerReputationCheckEvent |
评估玩家信誉时 |
ModuleToggleEvent |
模块开关状态改变时 |
ThreatScoreChangedEvent |
玩家威胁分数发生变化时 |
HoneypotTrapEvent |
IP 命中蜜罐时 |
IntelligenceAlertEvent |
威胁等级发生变化时 |
AttackSnapshotCompleteEvent |
取证报告生成完毕时 |
❓ 常见问题
这会降低我的服务器性能吗?
不会。AtomGuard 设计为轻量级。VPN 查找和日志记录等重型操作都在单独的线程上运行,不会影响服务器的性能。
它会屏蔽我的正常玩家吗?
非常不可能。VPN 系统需要 7 个提供商中的 2 个达成一致才会进行屏蔽,而机器人检测使用的是评分系统——并非简单的开/关切换。使用合法客户端的玩家将顺利通过且无感知。
它适用于破解版/离线模式服务器吗?
可以,但像 Mojang 账户验证等功能将不适用。无论在线还是离线模式,机器人检测、DDoS 防护和漏洞修复均可正常工作。
我可以只使用 Velocity 模块或仅使用 Core 插件吗?
完全可以。它们独立工作。同时使用两者可获得最大程度的保护,或者只使用你需要的那个。
如何更新?
只需替换 jar 文件并重启即可。AtomGuard 会自动将你的配置迁移到新版本,并为旧版本创建备份。
我遇到了误报——玩家被错误地屏蔽了。
尝试在配置文件中调整机器人检测阈值,或将受信任的 IP 添加到白名单中。默认设置已针对大多数服务器进行了优化,但每个社区的情况都不同。
我在哪里寻求帮助?
在 GitHub 上提交 Issue ——我们很乐意提供帮助!
Announcament!: Project Will Not Get Any Updates
AtomGuard
Keep your Minecraft server safe from hackers, bots, and crashes.
Works with Paper 1.21.4 and Velocity 3.x
AtomGuard is a free, open-source security plugin that protects your Minecraft server from attacks, exploits, and cheaters — so you can focus on building your community.
⬇️ Download · ? Setup Guide · ❓ FAQ · ? Contribute
? Why AtomGuard?
Running a Minecraft server is fun — until someone tries to ruin it. Bot attacks that flood your server with fake players, exploits that crash everyone out, duplication glitches that destroy your economy… the list goes on.
AtomGuard handles all of that for you. Just drop it in your plugins folder and your server is protected. No complicated setup, no networking degree required.
Here's what it protects you from:
?️ Bot & DDoS Attacks — Automatically detects and blocks waves of fake players and connection floods before they slow your server down.
? VPN & Proxy Abuse — Stops banned players from rejoining on a VPN. Uses 7 different detection methods for accuracy, and won't accidentally block your real players.
? Crash Exploits — Fixes 44+ known Minecraft exploits that hackers use to crash servers, including book crashes, NBT attacks, and packet exploits.
♊ Duplication Glitches — Prevents item duplication bugs that can wreck your server's economy.
⚡ Lag Machines — Limits redstone, explosions, pistons, and entities so players can't build devices that lag your server.
? Country Filtering — Allow or block players from specific countries if you want a region-locked community.
? Login Protection — Blocks brute-force password attacks and prevents the use of common passwords (works with AuthMe).
✨ Feature Highlights
? Smart Bot Detection
AtomGuard doesn't just look at one thing — it scores each connecting player across 8 different signals like connection speed, client type, join pattern, and username. Players that score low enough get through instantly. Suspicious ones get a quick math challenge. Obvious bots get blocked.
Your regular players won't even notice it's running.
⚔️ Adaptive DDoS Protection
The SmartThrottle Engine automatically adjusts its defenses based on how bad an attack is:
? Normal → ? Elevated → ? High → ? Critical → ⛔ Lockdown
During a serious attack, it locks down new connections while making sure your verified players can still get in. Once the attack stops, everything goes back to normal automatically.
? Threat Intelligence
AtomGuard learns your server's normal traffic patterns over 7 days. When something unusual happens — like a sudden spike in connections at 3 AM — it automatically ramps up protection. This means fewer false alarms and faster response to real threats.
? Player Trust Score
The more a player plays on your server without causing trouble, the more AtomGuard trusts them. Veteran players skip most checks entirely, which means faster logins for your loyal community and stricter checks for newcomers.
| Tier |
Who |
What Happens |
| ? New |
Just joined |
Full security checks |
| ? Regular |
Been around a while |
Standard checks |
| ✅ Trusted |
Clean track record |
Skips attack-mode checks |
| ⭐ Veteran |
Long-time player |
Skips bot & VPN checks too |
? Honeypot
AtomGuard can run a fake Minecraft server on a different port. When bot scanners find it and try to connect, their IP gets instantly banned from your real server. It's like a trap for attackers.
? Attack Reports
After every attack, AtomGuard saves a detailed report — what happened, when, how many IPs were blocked, and which defenses did the work. You can review these anytime to understand what your server faced.
? What You'll Need
| What |
Version |
Required? |
| ☕ Java |
21 or newer |
✅ Yes |
| ? Paper server (or a fork like Purpur) |
1.21.4 |
✅ Yes |
| ? PacketEvents plugin |
2.6.0+ |
✅ Yes (for the core plugin) |
| ? Velocity proxy |
3.x |
Only if you use a proxy |
| ?️ MySQL database |
8.0+ |
Optional — for persistent data |
| ? Redis |
7.x |
Optional — for multi-server sync |
| ? MaxMind license key |
— |
Only if you want country filtering |
? Getting Started
Option 1: Paper Server (most people start here)
Step 1 — Download PacketEvents and put it in your plugins/ folder.
Step 2 — Download AtomGuard-core.jar and put it in the same plugins/ folder.
Step 3 — Start (or restart) your server. AtomGuard will create its config files automatically.
Step 4 — Customize the settings in plugins/AtomGuard/config.yml if you want — but the defaults work great out of the box!
Option 2: Velocity Proxy
If you run a Velocity network, install the proxy module for network-wide protection:
Step 1 — Download AtomGuard-velocity.jar and put it in your proxy's plugins/ folder.
Step 2 — Start (or restart) your proxy. Config files are created automatically.
Step 3 — Edit plugins/atomguard-velocity/config.yml to your liking.
Step 4 (optional) — Enable the Redis section in both configs if you want the proxy and backend servers to share ban lists and threat data.
? Tip: You can run both modules together! The Velocity module stops threats at the network edge, while the Core module handles in-game exploits on each server.
? Commands
All commands require the atomguard.admin permission unless noted otherwise.
| Command |
What it does |
/atomguard status |
See which modules are running and their stats |
/atomguard reload |
Reload the config without restarting (needs atomguard.reload) |
/atomguard stats |
View the statistics dashboard |
/ag intel status |
Check current threat level |
/ag trust info <player> |
Look up a player's trust score |
/ag replay list |
Browse past attack reports |
/ag honeypot status |
Check if the honeypot is active |
/panic |
? Emergency lockdown — blocks ALL new connections (needs atomguard.panic) |
Permissions
| Permission |
What it does |
atomguard.admin |
Full access to all commands |
atomguard.bypass |
Player skips all security checks (give this to your staff) |
atomguard.notify |
Player receives in-chat alerts when exploits are blocked |
atomguard.reload |
Can reload the config (nothing else) |
atomguard.panic |
Can trigger emergency lockdown |
? How VPN Detection Works
Instead of relying on a single provider (which can be inaccurate), AtomGuard queries 7 providers at the same time and only blocks a player if at least 2 providers agree the IP is a VPN or proxy. This dramatically reduces false positives — your players on normal internet connections won't get blocked.
See the 7 providers
| # |
Provider |
How it checks |
| 1 |
Local Blocklist |
Your own custom IP list |
| 2 |
CIDR Blocker |
IP range rules |
| 3 |
DNSBL |
Spamhaus, DroneBL, and custom DNS blocklists |
| 4 |
IPHub |
Commercial VPN/proxy database |
| 5 |
ProxyCheck.io |
Real-time proxy detection API |
| 6 |
AbuseIPDB |
Checks IP abuse history |
| 7 |
IPApi |
Checks if the IP belongs to a hosting/datacenter provider |
Clean IPs are cached so returning players aren't checked again.
? Technical Details
These sections are for server admins who want to know exactly what's happening under the hood.
?️ Velocity Proxy Module — Full Details
The Velocity module intercepts connections at the proxy level before they ever reach your backend servers.
Connection Protection
- SYN Flood Detector — blocks IPs exceeding the connection threshold per second
- Slowloris Detector — identifies slow-drip connection drain attacks
- Traffic Anomaly Detector — Z-score, slow-ramp, and pulse attack detection
- Connection Fingerprinter — protocol + hostname + timing fingerprint to detect bot armies
- Subnet Analyzer — coordinated botnet detection at /24 and /16 level
- Sliding-Window Rate Limits — enforced at per-IP, per-subnet, and global levels
- Verified Player Shield — guarantees a slot for clean players during Critical/Lockdown
- Attack Session Recorder — full session log with JSON export
Bot Scoring Breakdown
| Signal |
Weight |
| Connection Speed |
20% |
| Join Pattern |
20% |
| Handshake Validity |
15% |
| Client Brand |
15% |
| Geo / Country |
10% |
| Username Pattern |
10% |
| Protocol Version |
10% |
Score thresholds: < 40 = pass, 40–60 = flagged, 60–75 = CAPTCHA challenge, 75–90 = kick, 90+ = auto-ban.
Additional layers: Brand analyzer (whitelists Fabric, Forge, Lunar, Badlion, LabyMod, OptiFine, Sodium), nickname blocker (regex patterns, length limits, special-character analysis), verified player cache (48-hour bypass for clean players).
Chat & Command Protection
Chat rate limiting, duplicate message detection, tab-complete flood blocking, command spam prevention, server-switch abuse prevention, packet size limits, crash loop detection (3+ disconnects in 30s), and short session flagging (< 3s).
Account Protection
IP reputation scoring (decays over time, rewards clean logins), auto-ban engine with grace periods, temporary ban manager, Mojang account verification, and hot-reloadable JSON blacklists/whitelists.
Kernel-Level Blocking (IPTables)
Can push bans directly to iptables, ip6tables, or nftables — blocked traffic never even reaches the JVM. Subnet banning at /24 level. Auto-cleanup on startup and shutdown.
Country / Geo Filtering
MaxMind GeoIP2 integration — whitelist or blacklist entire countries. Automatic weekly database updates.
Password Security (AuthMe)
Temporary ban after 5 failed logins, 10,000+ known weak passwords blocked, password similarity detection across the same IP.
? Core Plugin — Full Details
The core plugin runs on each Paper server and handles in-game security with 44+ modules.
Packet & Network Exploits
Invalid packet filtering at the Netty pipeline level, oversized packet blocking, offline packet injection prevention, packet timing & delay abuse detection.
NBT & Item Attacks
Nested NBT depth limiting, oversized NBT payload detection, bundle crash prevention, item sanitization on all inventory operations.
World & Chunk Crashers
Book & lectern exploit fix, map label crash fix, item frame crash fix, sign exploit prevention, chunk crash protection.
Duplication Fixes
Bundle duplication, inventory click duplication, cow & mule duplication, general dupe prevention engine.
Performance Limiters
Redstone circuit rate limiting, explosion limiter, piston limiter, falling block limiter, per-chunk entity limiter.
AtomShield™ Behavioral Analysis
Analyzes 9 signals per player: connection rate, gravity validation, packet timing, ping & handshake, protocol, username pattern, first-join behavior, post-join behavior, and heuristic profiling. Builds real-time behavioral profiles. Attack Mode auto-activates when TPS drops or connection floods are detected.
Threat Intelligence Engine
168-hour EMA traffic profile (7-day rolling baseline), Z-score anomaly detection across 3 threat levels, 3-minute confirmation window before escalation (prevents false positives), auto attack-mode on critical anomaly. Command: /ag intel <status|reset>
Player Trust Score
EMA-weighted formula: playtime + clean sessions + violation history. Persistent storage via trust-scores.json. Command: /ag trust <info|set|reset|top>
Forensic Analysis
Attack snapshots with UUID, timeline, peak rate, blocked IPs, and per-module stats. 4 severity levels (LOW/MEDIUM/HIGH/CRITICAL). Auto-export to forensics/attack-<uuid>.json. Command: /ag replay <list|latest|<id>|export>
Honeypot Module
Fake TCP Minecraft server (SLP protocol) that lures bot scanners. Auto-blacklists probing IPs. Command: /ag honeypot <status|stats>
⚡ Integrations
| Integration |
What it does |
| MySQL + HikariCP |
Persistent database storage with connection pooling |
| Redis Pub/Sub |
Sync bans, threat data, and trust scores across your entire network |
| Discord Webhooks |
Get instant Discord notifications when attacks happen |
| Web Panel |
Browser-based dashboard with live stats |
| Config Migration |
Automatic config upgrades when you update — with backups |
| Async Logging |
All logging happens off the main thread with 7-day rotation |
| Hot Reload |
Change settings without restarting your server |
? Developer API
Want to integrate with AtomGuard or build on top of it? There's a full API available.
See API documentation
Maven Dependency
<dependency>
<groupId>com.atomguard</groupId>
<artifactId>AtomGuard-api</artifactId>
<version>1.2.2</version>
<scope>provided</scope>
</dependency>
Quick Examples
// Check an IP's reputation score
IReputationService rep = AtomGuardAPI.getInstance().getReputationService();
int score = rep.getScore(player.getAddress().getAddress());
// Enable or disable a module at runtime
IModuleManager modules = AtomGuardAPI.getInstance().getModuleManager();
modules.setEnabled("bot-koruma", false);
// Access trust scores, forensics, and intelligence
AtomGuardAPI.getInstance().getTrustScoreManager();
AtomGuardAPI.getInstance().getForensicsManager();
AtomGuardAPI.getInstance().getIntelligenceEngine();
// Listen for events
@EventHandler
public void onExploitBlocked(ExploitBlockedEvent event) {
String module = event.getModuleName();
Player player = event.getPlayer();
}
Available Events
| Event |
When it fires |
ExploitBlockedEvent |
An exploit is blocked |
AttackModeToggleEvent |
Attack mode turns on or off |
PlayerReputationCheckEvent |
A player's reputation is evaluated |
ModuleToggleEvent |
A module is toggled |
ThreatScoreChangedEvent |
A player's threat score changes |
HoneypotTrapEvent |
An IP hits the honeypot |
IntelligenceAlertEvent |
Threat level changes |
AttackSnapshotCompleteEvent |
A forensic report is finalized |
❓ FAQ
Will this slow down my server?
Nope. AtomGuard is designed to be lightweight. Heavy operations like VPN lookups and logging happen on separate threads so your server's performance isn't affected.
Will it block my regular players?
Very unlikely. The VPN system requires 2 out of 7 providers to agree before blocking, and the bot detection uses a scoring system — not a simple on/off switch. Players using legitimate clients will pass through without noticing.
Does it work with cracked / offline-mode servers?
Yes, but some features like Mojang account verification won't apply. Bot detection, DDoS protection, and exploit fixes work regardless of online/offline mode.
Can I use just the Velocity module or just the Core plugin?
Absolutely. They work independently. Use both together for maximum protection, or just the one you need.
How do I update?
Just replace the jar file and restart. AtomGuard automatically migrates your config to the new version and creates a backup of the old one.
I'm getting false positives — players are being blocked incorrectly.
Try adjusting the bot detection threshold in your config, or add trusted IPs to the whitelist. The default settings are tuned for most servers, but every community is different.
Where do I get help?
Open an issue on GitHub — we're happy to help!
评论 0
登录后参与评分
还没有评论,来抢沙发吧