AccountShield
你的网络所缺失的账户完整性层。 每个网络都会因同一个敌人而浪费时间:小号账号。封禁规避、利用小号刷取经济资源、奖励滥用以及入侵机器人。AccountShield 会对每次登录进行指纹识别,并将其与它见过的所有账号进行比对得分——分数越高,是同一人的可能性就越大。
无需数据库。无需外部服务。插件旁边放一个普通的 YAML 文件即可。
它能检测什么
- 小号账号 - 提供带分数和人类可读原因的排名匹配结果。
- 封禁规避 - 与被封禁账号关联的登录会在最早阶段被阻止,并在加入服务器后使用更完整的指纹重新检查。
- 入侵机器人 / 加入洪水 - 针对每个网络范围设置的滑动窗口机制,在经典“来自同一 VPS 范围的 100 个机器人”入侵到达你的服务器列表之前将其拦截。
- 小号农场 - 让你知道哪些共享相同指纹的多个账号正在同时在线。
指纹工作原理
对于每个账号,它会记录 IP 地址和一个粗略的范围(IPv4 /24,IPv6 /64)、客户端品牌、区域设置和稳定的客户端设置(皮肤部件、主手、聊天选项等)、客户端的模组/插件渠道集,以及用户名。
信号会被加权并求和(上限为 100):
| 信号 |
权重 |
| 完全相同的 IP |
45 |
| 相同的客户端指纹 |
30 |
| 匹配的模组集(重叠度 >= 80%) |
20 |
| 相似的用户名 |
15 |
| 相同的范围(/24, /64) |
12 |
| 近期从相同范围活跃 |
10 |
| 相同的区域设置 |
3 |
仅凭共享 IP 不会将两个账号链接在一起——家庭用户可通过
/accountshield trust <a> <b> 进行处理。
命令
| 命令 |
描述 |
/alts <player> |
带分数和原因的小号排名匹配 |
/accountshield info <player> |
存储的关于某账号的所有信息 |
/accountshield ban / unban |
管理封禁 |
/accountshield trust / untrust |
将你确认为不同人的账号列入白名单 |
/accountshield watch <player> |
登录警报 |
/accountshield list |
计数 + 监控列表 |
别名:/ashield, /as, /shield。
警报
可向游戏内工作人员、控制台以及可选地发送到 Discord Webhook。
配置
阈值、每个信号的权重、名称/模组相似度、封禁规避操作(deny / kick / alert)、入侵窗口和限制、对原版封禁的尊重以及 Webhook 均位于 config.yml 中。
诚实的限制
检测是基于启发式的。VPN 会改变 IP 信号;全新的客户端会改变指纹。AccountShield 提高了开小号的成本并抓住了大多数粗心的人——它不是魔法。
AccountShield
The account-integrity layer your network is missing. Every network loses time to the same
enemy: alt accounts. Ban evasion, alt-farming the economy, reward abuse and raid bots. AccountShield
fingerprints every login and scores it against every account it has ever seen - the higher the score,
the more likely it is the same person.
No database. No external services. Plain YAML next to the plugin.
What it catches
- Alt accounts - ranked matches with scores and human-readable reasons.
- Ban evasion - a login linked to a banned account is blocked at the earliest point, and re-checked after join with the fuller fingerprint.
- Raid bots / join floods - a sliding window per network range stops the classic "100 bots from one VPS range" raid before it reaches your server list.
- Alt-farming - know when several accounts that share a fingerprint are active at once.
How the fingerprint works
For each account it records the IP and a coarse range (/24 IPv4, /64 IPv6), the client brand,
locale and stable client settings (skin parts, main hand, chat options...), the client's mod/plugin
channel set, and the username.
Signals are weighted and summed (capped at 100):
| Signal |
Weight |
| Same exact IP |
45 |
| Identical client fingerprint |
30 |
| Matching mod set (>= 80% overlap) |
20 |
| Similar username |
15 |
| Same range (/24, /64) |
12 |
| Recently active from same range |
10 |
| Same locale |
3 |
A shared IP alone will not link two accounts - households are handled with
/accountshield trust <a> <b>.
Commands
| Command |
Description |
/alts <player> |
Ranked alt matches with scores and reasons |
/accountshield info <player> |
Everything stored for an account |
/accountshield ban / unban |
Manage bans |
/accountshield trust / untrust |
Whitelist accounts you know are different people |
/accountshield watch <player> |
Login alert |
/accountshield list |
Counts + watchlist |
Aliases: /ashield, /as, /shield.
Alerts
In-game to staff, to console, and optionally to a Discord webhook.
Configuration
Thresholds, every signal weight, name/mod similarity, ban-evasion action
(deny / kick / alert), raid window and limits, vanilla-ban respect and the
webhook are all in config.yml.
Honest limits
Detection is heuristic. A VPN changes the IP signal; a fresh client changes the
fingerprint. AccountShield raises the cost of alting and catches the careless
majority - it is not magic.
评论 0
登录后参与评分
还没有评论,来抢沙发吧