Dash
Dash 是一款 Minecraft 服务器的自托管 Web 管理仪表盘。
无需在终端窗口、FTP 客户端、配置文件文件夹和互不关联的管理插件之间来回切换即可运行服务器。Dash 将实时性能、控制台访问、玩家管理、文件操作、备份、报告、权限、更新和恢复工具整合到一个响应式的浏览器工作区中。
Dash 构建为一组原生服务器仪表盘:
| 项目 |
用途 |
安装位置 |
| Dash |
Paper, Purpur, Spigot, Bukkit 及兼容服务器 |
plugins/ |
| FabricDash |
Fabric 服务器 |
mods/ |
| ForgeDash |
NeoForge 服务器 |
mods/ |
| NeoDash |
用于多服务器的可选中央控制平面 |
Linux 主机或 Docker |
每个本地仪表盘旨在管理其实际运行的服务器。NeoDash 是可选的:当您希望在多个通过 Dash 连接的服务器之间实现全局可见性、配置分发和签名交接时,请使用它。
5.0 新功能
Dash 5.0 是目前最大的版本发布:采用新的 Material 3 Expressive 界面(支持 14 种语言)、以标准 MCP 服务器取代内置 AI、更加谨慎的 Autopilot(自动修复)、完整的安全层以及为您运营服务器的运维和社区工具。
- Material 3 Expressive 界面:每页均配备,支持手机布局、深色和浅色主题以及 14 种语言
- Autopilot:解释清晰、受保护且可撤销的自动修复,附带学习统计数据和自定义规则
- AI 访问 (MCP):通过作用域 API 令牌连接 Claude、ChatGPT 兼容客户端、Cursor 或脚本,默认关闭
- 安全:全员双因素登录和密码密钥,带有新设备警报的会话概览,IP 白名单,签名审计日志,引导式密钥轮换
- 运维:带倒计时的计划重启,一键维护模式,延迟侦探,带试运行和回滚功能的更新管理器,带自动恢复测试的加密异地备份
- 社区:公共状态页,白名单申请,封禁申诉,聊天审核,GDPR 数据导出与擦除
- 不再处于测试阶段:Intelligence、Maintenance 和 Guardian 已成为常规功能;内部 Dash AI 已移除,由 MCP 取代
下表展示了各项目在 5.0 中包含的内容。FabricDash 和 ForgeDash 的加载器限制:更新管理器无法在模组服务器上运行 Shadow Boot Lab 中的试运行(它会提示这一点且从不报告通过;安装需要显式覆盖);Windows 上 Fabric 的更新交换仅在服务器进程结束后完成;重启需要启动器脚本或能重新启动服务器的宿主;加入消息中的应用和申诉提示仅限 Paper;聊天审核没有绕过权限(操作员可豁免),且 Fabric 无法重写消息,它会拒绝并重发被审查的文本作为系统行。
| 功能 |
Dash (Paper) |
FabricDash / ForgeDash |
NeoDash |
| Material 3 Expressive UI, 14 种语言 |
是 |
是 |
是 |
| 作用域 API 令牌和写入 API |
是 |
是 |
是 |
| MCP 服务器 (AI 访问) |
是 |
是 |
是 (舰队工具) |
| Autopilot 核心 |
是 |
是 |
舰队卡片 |
| Autopilot 学习统计数据、自定义规则、暂停 |
是 |
是 |
舰队警告 |
| 2FA, 密码密钥, 会话, 新设备警报 |
是 |
是 |
是 |
| IP 白名单, 签名审计日志, 密钥轮换 |
是 |
是 |
是 |
| 计划重启, 维护模式, 延迟侦探 |
是 |
是 |
不适用 |
| 更新管理器, 异地备份 |
是 |
是 (无试运行) |
不适用 |
| 状态页, 申请, 申诉, 聊天审核, GDPR |
是 |
是 |
不适用 |
Dash 为您提供什么
Dash 不是硬绑到 Minecraft 上的通用托管面板。它是一个理解实时服务器及其日常运营工作的仪表盘。
您可以使用它来:
- 在玩家发现问题之前查看服务器健康状况
- 通过浏览器运行命令并读取实时控制台
- 管理玩家、审核、库存和关联的 Dash 账户
- 浏览、编辑、上传、下载和安全地组织服务器文件
- 通过 Modrinth 发现兼容的插件或模组
- 创建验证过的备份,准备维护并检查恢复选项
- 为员工提供他们所需的确切权限,并在敏感操作背后保留审计轨迹
- 通过专用的公共表单接收玩家报告,而无需暴露管理仪表盘
- 通过 NeoDash 在不同服务器之间迁移,同时保持安全且原生适配加载器的工作流
该界面专为实际运营设计:紧凑的控制元素、可读的数据、响应式布局、键盘支持、有用的加载和错误状态、减少动画支持以及导航功能确保在当前页面准备好之前当前页面保持可见。
服务器概览和实时控制台
仪表盘让您立即掌握关键状态:
- 服务器运行时间和当前 Dash 版本
- TPS、MSPT、内存、CPU 和已加载区块
- 在线玩家数量和各世界活动情况
- 受权限保护的服务器启动、重启和停止控件
- 健康和资源使用的实时图表
- 快速访问实时控制台和常见管理操作
远程控制台专为基于浏览器的管理而构建,而不仅仅是一个静态日志查看器。
- 读取当前控制台输出
- 从专注的命令字段发送服务器命令
- 使命令历史在当前会话中可用
- 清除可见的控制台视图而不删除实际的服务器日志
- 通过服务端权限限制访问
对于历史指标,Graphs 提供真实的 1h、6h、24h 和 7d 范围,支持实时或暂停刷新、响应式画布、工具提示、空状态以及 JSON 或 CSV 导出。
玩家管理和玩家档案
Dash 将玩家管理转变为一个正式的工作空间,而不是一个长长的、水平滚动的表格。
- 面向桌面端和移动端的响应式玩家卡片
- 显示玩家名称、UUID、状态、延迟(ping)、所在世界及位置
- 皮肤头像与点击即可打开的交互式 3D 皮肤查看器
- 支持拖拽旋转和滚轮缩放皮肤预览;若无法渲染 3D 则提供静态备用方案
- 玩家档案、会话历史记录及工作人员备注
- 在权限允许的情况下,执行踢出(kick)、封禁(ban)、解封(unban)和加入白名单等操作的管理控制
- 在活动平台支持时,提供物品栏和末影箱工具
- 基于权限的操作界面,而非千篇一律的管理员面板
- 直接在玩家记录中显示现有的 Dash 账户链接,并方便执行允许的账户操作
关联账户视图对团队至关重要:操作员可以确认某个 Minecraft 身份是否已连接到某位 Dash 用户,而无需在两个独立的系统之间猜测。
文件、插件和模组
文件管理器有意设计为具备服务器感知能力且受到严格保护。它为工作人员提供实用的控制功能,但不会将服务器根目录当作无限制的文件浏览器处理。
- 整行进行文件和文件夹交互:点击文件气泡,而非微小的文件名目标区域
- 浏览文件夹、配置文件、日志、世界、插件或模组
- 上传文件时无需将页面滚动回顶部
- 在当前角色权限允许的情况下,创建、编辑、重命名、下载和删除文件
- 将文件夹打包下载为 ZIP 归档文件
- 检测二进制文件,并从不安全操作中保护锁定文件或 PID 文件
- 在界面背后使用规范路径处理、有上限的上传及安全归档规则
本地“插件”或“模组”工作区与 Modrinth 发现功能协同工作:
- 查看已安装的插件或模组及其本地状态
- 当服务器策略允许时,上传本地 JAR 文件
- 从仪表板搜索 Modrinth
- 按当前服务器加载器和精确的 Minecraft 版本筛选结果
- 解析适合正确平台和补丁的可安装文件,而不是显示不相关的项目
- 在 Dash、FabricDash 或 ForgeDash 中保持本地加载器的维护,确保服务器上下文准确无误
工单与公开玩家举报
Dash 包含两种相互关联但刻意分离的工作流程:Dash 内部的工作人员工单和外部公开的玩家举报。
工作人员工单
工单工作区包括:
- 收件箱、我的工单、新建举报、工作人员备注和举报链接视图
- 搜索、筛选、分页、状态和优先级
- 回复及内部工作人员上下文信息
- 基于权限的状态更改和删除操作
- 清晰的空状态、加载状态和验证状态提示
无需仪表板访问权限的公开举报
每位玩家都可以创建举报,而无需成为 Dash 管理员。
/dash ticket 创建举报链接
/dash report <player> 准备玩家举报
- 链接具有时效性、单次使用性质,并打开专用的纯举报网站
- 举报页面永远不会变成仪表板登录界面,也不会暴露管理员路由
- 举报可附加有限的自动上下文信息,如举报者 UUID、世界、坐标、延迟和服务器版本
- 公开 URL 使用配置的 HTTPS
report-url,因此可以使用服务器的真实 SSL 域名,而不是私有 IP 或原生 HTTP 地址
举报归属于本地的 Dash、FabricDash 或 ForgeDash 服务器。NeoDash 不会将未认证的举报用户重定向到集群面板。
用户、角色、权限与审计
Dash 拥有自己的仪表板用户系统,供服务器团队使用。
- 创建用户、邀请操作员并分配基于角色的访问权限
- 使用内置的等级体系或配置自定义角色
- 保护主管理员(Main Admin)账户及其敏感控制功能
- 通过签名桥接工作流程处理 NeoDash 关联的用户
- 将访问范围限定于仪表板数据、控制台、服务器控制、玩家操作、物品栏工具、文件、插件或模组、用户、设置、备份、举报、通知及高级工作区
- 在审计日志中追踪重要的管理操作
审计日志为以下事件提供有用的操作证据:认证、命令、文件更改、设置更改、插件或模组操作、备份工作流、举报操作以及批准的高风险更改。
备份、维护与更新
Dash 为操作员提供了一种更为审慎的方式来变更和恢复服务器。
- 创建经过验证的备份,并设置内容范围和保留策略控制
- 计划备份和定期服务器任务,例如
save-all
- 保留还原点,以便在需要恢复路径时应对更改
- 在执行高风险操作之前预览维护工作和验证备份
- 运行恢复演练、配置漂移检查、容量预测和兼容性检查
- 审查已安装的扩展状态、更新信息及支持的更新路径
- 使用基于摘要(digest)的下载方式,而不只是依赖文件名
维护工作保持在服务器本地。集群范围的规划、配置和生命周期控制属于 NeoDash 的职责;特定加载器的服务器工作则留在拥有该服务器的仪表板中进行。
值得信赖的操作
这些工具接管了围绕重启、更新和备份的日常例行工作。在你开启它们之前,它们处于关闭或最安全模式,且所有操作均会被审计。
带倒计时的计划重启
- 支持每日、每周或一次性计划,并考虑时区和夏令时规则
- 在聊天中以标题形式向玩家发出警告,你可以在指定时间点触发(例如提前 10 分钟、1 分钟、10 秒)
- 可选静默窗口:重启会等待在线玩家,并在你设定的时限后强制执行
- 手动跳过、取消或启动倒计时;控制台命令为
/dash restart schedule|cancel|now
- 与维护窗口相关联,并在仪表板卡片和状态页面上显示
一键维护模式
- 开启白名单,在服务器列表中设置维护消息,并在倒计时结束后礼貌地踢出玩家
- 管理员不受影响;关闭维护模式后,之前的白名单状态和每日公告(MOTD)会自动恢复
- 可选自动结束时间、每个面板页面显示横幅,以及支持重启后保持状态的机制;使用命令
/dash maintenance on|off|status
卡顿侦探
- 通过提供证据回答“为什么现在卡顿?”:TPS 和 MSPT 历史记录、区块、实体和任务
- 如果安装了 spark,它会运行简短的性能分析并将开销归因于插件;如果没有安装 spark,则会提示并采用清晰的启发式方法进行分析
- 对嫌疑对象进行排名,提供通俗易懂的解释和建议的操作,且绝不会自行更改任何内容
带试运行和回滚功能的更新管理器
- 通过 Modrinth(文件哈希查找)和 GitHub 查找更新,提供更新日志、版本类型和兼容性信息
- 首先在 Shadow Boot Lab 中的服务器副本上测试更新,并提供带有证据的通过或失败报告
- 仅在测试通过(或手动覆盖)后才部署,保留旧文件和配置的备份
- 监控重启后的最初几分钟,如果在出现错误、插件被禁用、TPS 下降或重启循环时会自动回滚
- 支持 30 天内的手动回滚;仅通过 HTTPS 从 Modrinth 或 GitHub 下载并进行哈希验证;默认情况下仅为通知模式
加密的异地备份与自动恢复测试
- 将备份上传至兼容 S3 的存储(AWS、Backblaze B2、Wasabi、MinIO)、Google Drive 或本地/NAS 文件夹
- 数据在离开服务器前已在服务器上通过 AES-256-GCM 进行加密;密钥由您自行保管
- 提供独立的
Decrypt.java,可在不依赖插件的情况下恢复备份
- 可选每周恢复测试:下载最新副本,解密并将其解压到临时文件夹中进行检查,不会影响正在运行的服务器
- 支持保留策略、重试队列、带宽限制和故障警报
通知与投递健康度
通知围绕实际的目标地址和投递状态组织,而非无效的占位符字段。
- 当前通知状态概览
- 目标地址配置与测试投递
- 针对相关服务器和安全信号的事件选择
- 投递健康状况与可见的失败反馈
- 在已配置的情况下,支持按用户或角色感知的通知控制
- 保留现有的 Discord 投递功能作为真正的服务器后端集成
情报中心
情报中心是为需要超越单一健康数值的操作员提供的先进可选工作区。它围绕真实事故中产生的问题组织:发生了什么、什么可以安全地更改、谁受到影响以及如何恢复?
调查
- 具有受限输入和安全失败结果的 Shadow Boot Lab
- 根本原因探索器
- 性能回归雷达
- 插件或模组资源归属分析
- 引导式安全模式
- 依赖关系与影响地图
变更与恢复
- 状态时光机
- 备份内容浏览器
- 自适应维护窗口
- 操作护栏
- 恢复演练与恢复预览
- 配置漂移与容量预测
- 加载器感知的扩展检查兼容性中心
关怀玩家与管理员
- 玩家旅程回放
- 支持与申诉收件箱
- 跨服务器玩家身份
- 玩家体验评分
- 即时管理员访问权限
- 双重控制操作
保护服务
- 保留中心
- 供应链中心
- 配置 Schema 助手
- 通用搜索
- 世界健康中心
- 存储智能
- 服务级别仪表板
- 操作员作战室
- 公共状态页
情报中心是常规功能之一。可以在设置中禁用它,而无需移除标准仪表板。
Guardian
Guardian 将安全证据、调查和恢复集中在一个专注的工作区中,而不是冗长的页面。
- 调查事件证据及相关历史记录
- 处理案件、管理员笔记和针对性搜索
- 审查保护策略、活动记录和警报
- 使用恢复预览以及受控的回滚或恢复操作
- 配置保留策略,同时使破坏性操作受到权限限制
- 分页显示长活动流,以保持事件时间线的可用性
Guardian 是可选功能,可以独立禁用。高影响操作仍需经过服务器授权、预览和审计。
Autopilot:谨慎的自动修复
Autopilot 读取您的日志,找出问题的原因并提出最小的安全修复方案。您可以决定赋予它多大的自主权:
- 关闭、建议(默认,仅显示它将执行的操作)、批准(等待您点击确认)或 自动(在严格限制内独立行动)
- 每次发现都会显示其置信度、背后的因素、证据行、替代方案以及可能改变其判断的条件
- 护栏机制:每日操作上限、冷却时间、免打扰时段、黑名单,以及仅在您允许时才进行重启
- 所有操作都会被记录,事后进行验证(“问题真的解决了吗?”),并且可以撤销
- 学习统计:Autopilot 正确的频率、您拒绝或撤销的频率,按问题类型分类,并提供一个仅由您应用的建议置信度阈值
- 自定义规则:“如果出现此错误,则通知我”或“则提出此操作”,并配有测试按钮;规则的执行范围永远不会超过护栏所允许的范围
- 拒绝即静默:被拒绝的建议不会再次出现,直到证据恶化或您设定的暂停期(1小时至永久)结束
- NeoDash 增加 集群警告:如果一个插件在一台服务器上造成麻烦,它会检查运行相同插件的其他所有服务器,并提前向您发出警告
随团队成长的安全机制
- 每个用户的两步验证:支持认证器应用(本地生成二维码,无需外部服务)、十个一次性恢复码,以及适用于所有用户或指定角色的策略,并附带宽限期
- Passkeys(通行密钥):使用指纹、手机或安全密钥登录;通行密钥可替代密码加验证码的组合
- 会话与设备管理:查看当前登录位置,注销单个设备,并在检测到新设备时收到警报;管理员可查看所有人的会话状态
- IP 白名单:将面板访问限制在您的网络范围内(支持 IPv4 和 IPv6),具备可信代理(trusted-proxy)支持、“我会把自己锁在外面吗?”的预览功能,以及用于紧急情况的控制台开关(
/dash allowlist off)
- 签名审计日志:每条记录都链接到前一条并进行定期签名;您可以导出日志并使用内置验证器在面板内或离线验证是否被篡改
- 引导式密钥轮换:自动替换 Dash 与 NeoDash 之间的桥接密钥,包含连接测试和回滚机制;重新测试 Discord Webhook,并查看哪些 API 令牌需要轮换
- 作用域 API 令牌:每个令牌仅携带您赋予的作用域权限,且绝不超过其拥有者的权限范围
社区工具
- 专属公共网站。启用公共站点和状态页面,对应用进行白名单管理;申诉、工单和玩家举报等服务部署在独立端口上,不暴露其他内容:在该端口上,登录页、面板、API 和管理员页面均返回 404。您可以使用 nginx、Caddy 或 Cloudflare 等将域名指向该地址,例如
https://paper.example.de/whitelist。主管理员可在面板中编辑每条路径、站点标题和公共基础 URL,更改即时生效。工作人员页面的仪表盘卡片及复制/打开按钮会显示站点地址,执行 /dash site 命令即可打印地址。在 Paper 平台上,未列入白名单的玩家会在踢出消息中看到申请地址,被封禁的玩家则可以看到申诉地址。
- 重启按钮:当 Dash 链接到 NeoDash 时,重启操作不再卡在“正在重启”状态(浏览器现在会打开 NeoDash 的确认页面);NeoDash 自身的重启和停止控制针对当前服务器运行;在没有重启脚本的情况下,Dash 会诚实地告知服务器仅会停止。NeoDash 集群:打开服务器界面速度更快,不再需要多次点击。卡顿侦探:诊断运行期间页面不再每几秒刷新一次。
以上内容默认关闭,仅主管理员可以开启。
- 公共状态页面:实时服务器状态、24 小时、7 天和 30 天的在线时长、事故通知和计划维护、JSON 数据源及可嵌入徽章;每个组件均为可选启用
- 白名单申请:带有反滥用保护的公开表单,工作人员审核,一键批准并为申请人提供状态链接
- 封禁申诉:被封禁的玩家可以在不了解其他玩家姓名的情况下发起申诉;工作人员可查看封禁详情,并在冷却期后选择解除封禁或拒绝
- 聊天审查:能够识别如 Leetspeak 和空格变体等规避技巧的词过滤系统、链接保护、反垃圾规则、渐进式静音阶梯以及可疑玩家概览,数据保留时间由您决定
- GDPR 合规工具:将 Dash 所知悉的所有玩家信息导出为 ZIP 文件,或通过预览和手动输入确认进行擦除;审计链保持完整,收据经过签名
AI 访问:为您自己的助手提供标准 MCP 服务器
Dash 不再自带 AI 功能。相反,它通过开放的模型上下文协议(MCP)进行交互,因此您已使用的助手(Claude、兼容 ChatGPT 的客户端、Cursor、脚本等)可以查看并在您授权的情况下操作您的服务器。此功能默认关闭;仅主管理员可在 AI 访问 设置中开启。
- 流式 HTTP 端点
POST /mcp(JSON-RPC 2.0),提供工具只读资源
- 通过 API 访问 的作用域 API 令牌进行身份验证:令牌以其拥有者的身份行事,且绝不超过拥有者的权限范围
- 读取类工具:状态、性能历史、控制台、玩家、个人资料、Guardian 查询、Intelligence 诊断、Autopilot 状态、备份、插件、文件(面板数据、密钥和机密信息永远不可读取)、审计日志、工单
- 操作类工具:控制台命令(受 API 拒绝列表和安全护栏约束)、踢出、封禁、解封、加入白名单(必须填写理由)、Autopilot 审批、备份、工单
- 每个工具都会标明是只读还是具有破坏性;每个工具均可单独关闭;一键终止开关可同时切断所有操作
- 所有输出中的机密信息、令牌和地址均会被掩码处理,结果大小受限,调用频率受到限制,且每次调用都会写入审计日志
- 除非来源地址在允许列表中,否则拒绝浏览器页面请求(防止跨站攻击和 DNS 重绑定攻击)
- 提供复制粘贴的连接代码片段,以及页面上的预设选项“MCP 只读”和“MCP 操作员”
确定性崩溃和日志诊断功能仍可在维护模式中使用。
专为团队设计的仪表盘
首次运行设置提供三种合理的初始配置档案:
- 轻量版:仪表盘、控制台、玩家、文件、扩展、更新和基本设置
- 标准版(推荐):轻量版加上工单、通知、图表、用户、审计和插件设置
- 完整版:标准版加上所有高级管理工作区
Intelligence、Maintenance 和 Guardian 可通过功能可用性设置单独开启或关闭。
侧边栏可直接在使用处自定义:
- 通过鼠标拖拽或触摸长按交互移动项目
- 调整条目顺序以匹配团队的工作流程
- 隐藏未使用的项目
- 从自定义列表中恢复隐藏的项目
- 在多次访问之间保留所选的本地排序
通过用户语言选择器,可使用英语、德语、西班牙语及其他界面语言。
NeoDash:可选的多服务器管理
NeoDash 是 Dash 生态系统的中央控制面板。当您运营多台服务器或希望拥有一个专用的 Linux 托管管理界面时,请安装它。
NeoDash 代码仓库:
使用 NeoDash,你可以:
- 在一个地方查看已连接的服务器和集群健康状态
- 启动、停止、重启、创建、安装和扫描服务器目录
- 配置 Paper、Fabric、Quilt、NeoForge 和 Vanilla 服务器
- 选择 Minecraft 1.21 至 1.21.11、26.2 或 26.3,并搭配相应的 Java 运行环境
- 根据加载器和确切的 Minecraft 版本,从 Modrinth 解析正确的稳定版 Dash 构件
- 管理集群用户、角色、分配、审计历史、组和更新
- 运行经过验证的备份、作用域恢复以及可选的 Google Drive 镜像同步
- 通过签名且防重放的 SSO 打开 Dash、FabricDash 和 ForgeDash
- 将插件仪表板返回链接保持在配置的公共 HTTPS NeoDash 域名上
NeoDash 是可选的。单个服务器可以独立完美地运行 Dash、FabricDash 或 ForgeDash。
NeoDash 快速安装
curl -sSL https://raw.githubusercontent.com/Framepersecond/NeoDash/main/install.sh | bash
推荐使用原生 Linux。也提供 Docker 构件。当 NeoDash 直接管理 Minecraft 26.2 或 26.3 工作负载时,需要 Java 25。
兼容性与安装
选择与服务器加载器和确切 Minecraft 版本相匹配的构建。
| 服务器 |
构件 |
运行时 |
| Paper, Purpur, Spigot, Bukkit 1.21.x |
dash-*.jar |
Java 21 |
| Paper, Purpur, Spigot, Bukkit 26.2 |
dash-*.jar |
Java 25 |
| Paper, Purpur, Spigot, Bukkit 26.3 |
dash-*.jar |
Java 25 |
| Fabric 1.21 至 1.21.11 |
匹配的 fabricdash-mc1.21.x-*.jar |
Java 21 和 Fabric Loader 0.19.5+ |
| Fabric 26.2 |
fabricdash-mc26.2-*.jar |
Java 25 和 Fabric Loader 0.19.5+ |
| Fabric 26.3 |
fabricdash-mc26.3-*.jar |
Java 25 和 Fabric Loader 0.19.5+ |
| NeoForge 1.21 至 1.21.11 |
匹配的 forgedash-mc1.21.x-*.jar |
Java 21 |
| NeoForge 26.2 |
forgedash-mc26.2-*.jar |
Java 25 |
| NeoForge 26.3 |
forgedash-mc26.3-*.jar |
Java 25 |
Paper, Spigot, Purpur 和 Bukkit
- 下载 Dash。
- 将 JAR 文件放入
plugins/ 目录。
- 重启服务器。
- 以管理员身份运行
/dash register。
- 打开一次性设置链接并创建第一个管理员。
Fabric 和 NeoForge
- 下载针对确切 Minecraft 版本构建的构件。
- 将其放入
mods/ 目录。
- 重启服务器。
- 以管理员身份运行
/dash register。
- 完成一次性设置。
安全须知
Dash 功能强大,因为它可以控制真实的服务器。请像对待控制台或主机访问一样谨慎对待仪表板。
- 使用 HTTPS 或可信的反向代理
- 在分享报告链接之前,将
report-url 设置为公共 HTTPS 仪表板源地址
- 使用强且唯一的凭据,并保护主管理员账户
- 为每个操作员授予最小可行的角色权限
- 审查审计日志和通知传递健康状况
- 将控制台、文件、更新、备份和高级修改权限限制在受信任的员工范围内
- 给助手授予完成工作所需的最小令牌范围(从“MCP 只读”开始)
- 为每个账户开启双因素登录,并将审计日志公钥副本保存在服务器之外
Dash 强制执行认证会话、CSRF 保护、服务端权限、规范路径检查、有界请求和上传、受保护的文件规则、安全的归档处理、摘要感知的下载和操作审计日志。敏感密钥材料保留在服务端,并在界面中隐藏。
合作伙伴
Dash 很荣幸与 EmeraldHost(一家 Minecraft 服务器托管商)合作。使用代码 Frxme10 可获得服务器 10% 折扣。
如果 Dash 为你节省了时间,你可以在 Buy Me a Coffee 上支持其开发。谢谢!
项目
Dash 为单个服务器提供了严肃的 Web 管理界面。当该服务器成为集群时,添加 NeoDash。
Dash
Dash is a self-hosted web admin dashboard for Minecraft servers.
Run a server without juggling terminal windows, FTP clients, config folders and disconnected admin plugins. Dash puts live performance, console access, player care, files, backups, reports, permissions, updates and recovery tools into one responsive browser workspace.
Dash is built as a family of native server dashboards:
| Project |
Use it for |
Install location |
| Dash |
Paper, Purpur, Spigot, Bukkit and compatible servers |
plugins/ |
| FabricDash |
Fabric servers |
mods/ |
| ForgeDash |
NeoForge servers |
mods/ |
| NeoDash |
Optional central control plane for multiple servers |
Linux host or Docker |
Each local dashboard is designed to manage the server it actually runs on. NeoDash is optional: use it when you want fleet-wide visibility, provisioning and signed handoff between multiple Dash-connected servers.
New in 5.0
Dash 5.0 is the biggest release so far: a new Material 3 Expressive interface in 14 languages, a standard MCP server instead of a built-in AI, a much more careful Autopilot, a complete security layer, and operations and community tools that run your server for you.
- Material 3 Expressive interface on every page, with phone layout, dark and light themes and 14 languages
- Autopilot: explained, guarded, undoable automatic fixes with learning statistics and your own rules
- AI access (MCP): connect Claude, ChatGPT-compatible clients, Cursor or scripts through scoped API tokens, off by default
- Security: two-factor login and passkeys for everyone, session overview with new-device alerts, IP allow-list, signed audit log, guided secret rotation
- Operations: scheduled restarts with countdown, one-click maintenance mode, lag detective, update manager with trial run and rollback, encrypted offsite backups with automatic restore test
- Community: public status page, whitelist applications, ban appeals, chat moderation, GDPR export and erase
- No more beta: Intelligence, Maintenance and Guardian are regular features; the internal Dash AI is gone, replaced by MCP
The table shows what each project contains in 5.0. Loader limits on FabricDash and ForgeDash: the update manager cannot run a trial in the Shadow Boot Lab on a mod server (it says so and never reports a pass; installing needs an explicit override); the update swap on Windows with Fabric completes only after the server process ends; a restart needs a launcher script or host that starts the server again; the apply and appeal hints in the join message are Paper only; chat moderation has no bypass permission (operators can be exempt) and Fabric cannot rewrite a message, it refuses it and sends the censored text as a system line.
| Feature |
Dash (Paper) |
FabricDash / ForgeDash |
NeoDash |
| Material 3 Expressive UI, 14 languages |
yes |
yes |
yes |
| Scoped API tokens and write API |
yes |
yes |
yes |
| MCP server (AI access) |
yes |
yes |
yes (fleet tools) |
| Autopilot core |
yes |
yes |
fleet cards |
| Autopilot learning statistics, custom rules, snooze |
yes |
yes |
fleet warnings |
| 2FA, passkeys, sessions, new-device alerts |
yes |
yes |
yes |
| IP allow-list, signed audit log, secret rotation |
yes |
yes |
yes |
| Scheduled restarts, maintenance mode, lag detective |
yes |
yes |
not applicable |
| Update manager, offsite backups |
yes |
yes (no trial run) |
not applicable |
| Status page, applications, appeals, chat moderation, GDPR |
yes |
yes |
not applicable |
What Dash Gives You
Dash is not a generic hosting panel bolted onto Minecraft. It is a dashboard that understands a live server and its day-to-day operating work.
Use it to:
- See server health before players notice a problem
- Run commands and read the live console from a browser
- Manage players, moderation, inventories and linked Dash accounts
- Browse, edit, upload, download and safely organize server files
- Discover compatible plugins or mods through Modrinth
- Create verified backups, prepare maintenance and inspect recovery options
- Give staff exactly the permissions they need, with an audit trail behind sensitive actions
- Receive player reports through a dedicated public form without exposing the admin dashboard
- Move from one server to another through NeoDash without losing a secure, loader-native workflow
The interface is designed for real operations: compact controls, readable data, responsive layouts, keyboard support, useful loading and error states, reduced-motion support and navigation that keeps the current page visible until the next one is ready.
Server Overview and Live Console
The Dashboard gives you an immediate view of the state that matters:
- Server uptime and current Dash version
- TPS, MSPT, memory, CPU and loaded chunks
- Online player count and per-world activity
- Server start, restart and stop controls, protected by permissions
- Live charts for health and resource use
- Quick access to the live console and common administration actions
The remote console is built for browser-based administration, not just a static log viewer.
- Read current console output
- Send server commands from a focused command field
- Keep command history available for the session
- Clear the visible console view without deleting actual server logs
- Restrict access with server-side permissions
For historical metrics, Graphs provides real 1h, 6h, 24h and 7d ranges, live or paused refresh, responsive canvases, tooltips, empty states and JSON or CSV export.
Player Management and Player Profiles
Dash turns player administration into a proper workspace rather than a long, horizontally scrolling table.
- Responsive player cards for desktop and mobile
- Player names, UUIDs, status, ping, world and location
- Skin avatars and a click-to-open interactive 3D skin viewer
- Drag-to-rotate and scroll-to-zoom skin previews with a static fallback when 3D rendering is unavailable
- Player profiles, session history and staff notes
- Moderation controls for kick, ban, unban and whitelist actions where permitted
- Inventory and ender-chest tools where the active platform supports them
- Permission-aware actions instead of a one-size-fits-all admin surface
- Existing Dash-account links shown directly on the player record, with permitted account actions close by
The linked-account view matters for teams: an operator can see whether a Minecraft identity is already connected to a Dash user rather than having to guess from two separate systems.
Files, Plugins and Mods
The file manager is intentionally server-aware and guarded. It gives staff the useful controls without treating the server root like an unrestricted file browser.
- Whole-row file and folder interaction: click the file bubble, not a tiny filename target
- Browse folders, configuration, logs, worlds, plugins or mods
- Upload without throwing the page back to the top
- Create, edit, rename, download and delete files when the current role permits it
- Download folders as ZIP archives
- Detect binary files and preserve protected lock or PID files from unsafe actions
- Use canonical path handling, bounded uploads and safe archive rules behind the interface
The local Plugins or Mods workspace works together with Modrinth discovery:
- See installed plugins or mods and their local status
- Upload local JAR files when the server policy allows it
- Search Modrinth from the dashboard
- Filter results by the active server loader and exact Minecraft version
- Resolve an installable file for the correct platform and patch instead of showing unrelated projects
- Keep local loader maintenance in Dash, FabricDash or ForgeDash where the server context is exact
Tickets and Public Player Reports
Dash includes two connected but deliberately separate workflows: staff tickets inside Dash and public player reports outside it.
Staff tickets
The Tickets workspace includes:
- Inbox, My Tickets, New Report, Staff Notes and Report Links views
- Search, filters, pagination, statuses and priorities
- Replies and internal staff context
- Permission-aware status changes and deletion
- Clear empty, loading and validation states
Public reports without dashboard access
Every player can create a report without becoming a Dash administrator.
/dash ticket creates a report link
/dash report <player> prepares a player report
- Links are short-lived, single-use and open a dedicated report-only website
- The report page never becomes a dashboard login screen and does not expose admin routes
- Reports can attach bounded automatic context such as reporter UUID, world, coordinates, ping and server version
- The public URL uses the configured HTTPS
report-url, so it can use the server's real SSL domain instead of a private IP or native HTTP address
Reports belong to the local Dash, FabricDash or ForgeDash server. NeoDash does not redirect an unauthenticated report user into the fleet panel.
Users, Roles, Permissions and Audit
Dash has its own dashboard user system for server teams.
- Create users, invite operators and assign role-based access
- Use built-in rank levels or configure custom roles
- Protect the Main Admin account and its sensitive controls
- Handle NeoDash-linked users through the signed bridge workflow
- Scope access to dashboard data, console, server controls, player actions, inventory tools, files, plugins or mods, users, settings, backups, reports, notifications and advanced workspaces
- Track important administrative actions in the audit log
The audit log provides useful operational evidence for events such as authentication, commands, file changes, settings changes, plugin or mod actions, backup workflows, report actions and approved high-risk changes.
Backups, Maintenance and Updates
Dash gives operators a more deliberate way to change and recover a server.
- Create verified backups with scoped content and retention controls
- Schedule backups and recurring server tasks such as
save-all
- Keep restore points available for changes that need a recovery path
- Preview maintenance work and validate backups before high-risk actions
- Run restore drills, configuration-drift checks, capacity forecasts and compatibility checks
- Review installed extension state, update information and supported update paths
- Use digest-aware downloads rather than trusting a filename alone
Maintenance remains server-local. Fleet-wide planning, provisioning and lifecycle control belongs in NeoDash; loader-specific server work stays in the dashboard that owns the server.
Operations You Can Trust
These tools take over the routine work around restarts, updates and backups. They are off or in the safest mode until you switch them on, and every action is audited.
Scheduled restarts with countdown
- Daily, weekly or one-off schedules with time zone and daylight-saving rules
- Player warnings in chat and as titles at marks you choose (for example 10 minutes, 1 minute, 10 seconds)
- Optional quiet window: the restart waits while players are online and is forced after a limit you set
- Skip, cancel or start a countdown by hand;
/dash restart schedule|cancel|now for the console
- Linked to the maintenance window and shown as a dashboard card and on the status page
One-click maintenance mode
- Turns the whitelist on, sets a maintenance message in the server list and kicks players politely after a countdown
- Staff stay in; the previous whitelist state and MOTD come back when you switch it off
- Optional automatic end time, a banner on every panel page and a state that survives restarts;
/dash maintenance on|off|status
Lag detective
- Answers "why is it lagging right now?" with evidence: TPS and MSPT history, chunks, entities and tasks
- With spark installed it runs a short profile and attributes the cost to plugins; without spark it says so and uses clear heuristics
- Ranks suspects with a plain-language explanation and a recommended action, and never changes anything by itself
Update manager with trial run and rollback
- Finds updates through Modrinth (file hash lookup) and GitHub, with changelog, version type and compatibility
- Tests the update first on a copy of your server in the Shadow Boot Lab and reports pass or fail with evidence
- Rolls out only after a pass (or a typed override), keeps a backup of the old file and configuration
- Watches the first minutes after the restart and rolls back automatically on errors, a disabled plugin, a TPS drop or a restart loop
- Manual rollback for 30 days; downloads only from Modrinth or GitHub over HTTPS with hash verification; notify-only by default
Encrypted offsite backups with automatic restore test
- Upload backups to S3-compatible storage (AWS, Backblaze B2, Wasabi, MinIO), Google Drive or a local or NAS folder
- Encrypted on your server with AES-256-GCM before anything leaves it; the key stays with you
- A standalone
Decrypt.java restores a backup without the plugin
- Optional weekly restore test: downloads the newest copy, decrypts it, unpacks it into a temporary folder and checks it, without touching the live server
- Retention, retry queue, bandwidth limit and failure alerts
Notifications and Delivery Health
Notifications are organized around actual destinations and delivery state rather than inactive placeholder fields.
- Overview of current notification state
- Destination configuration and test delivery
- Event selection for relevant server and security signals
- Delivery health and visible failure feedback
- Per-user or role-aware notification controls where configured
- Existing Discord delivery retained as a real server-backed integration
Intelligence Center
Intelligence is the advanced optional workspace for operators who need more than a health number. It is organized around the questions that arise during real incidents: what happened, what can safely change, who is affected and how do we recover?
Investigate
- Shadow Boot Lab with bounded input and safe failure results
- Root Cause Explorer
- Performance Regression Radar
- Plugin or mod Resource Attribution
- Guided Safe Mode
- Dependency and Impact Map
Change and recover
- State Time Machine
- Backup Content Explorer
- Adaptive Maintenance Window
- Action Guardrails
- Restore drills and recovery previews
- Configuration Drift and Capacity Forecasting
- Compatibility Center for loader-aware extension checks
Care for players and staff
- Player Journey Replay
- Support and Appeals Inbox
- Cross-Server Player Identity
- Player Experience Score
- Just-in-Time Staff Access
- Dual-Control Actions
Protect the service
- Retention Center
- Supply Chain Center
- Config Schema Assistant
- Universal Search
- World Health Center
- Storage Intelligence
- Service Level Dashboard
- Operator War Room
- Public Status Page
Intelligence is a regular feature. It can be disabled in Settings without removing the standard dashboard.
Guardian
Guardian keeps security evidence, investigation and recovery in a focused workspace instead of an endless page.
- Investigate event evidence and nearby history
- Work with cases, staff notes and targeted search
- Review protection policies, activity and alerts
- Use recovery previews and guarded rollback or restore actions
- Configure retention while keeping destructive work permission-gated
- Page long activity streams so event timelines remain usable
Guardian is optional and can be disabled independently. High-impact actions stay server-authorized, previewed and audited.
Autopilot: Careful Automatic Fixes
Autopilot reads your logs, finds the cause of a problem and proposes the smallest safe fix. You decide how much freedom it gets:
- Off, Suggest (default, only shows what it would do), Approve (waits for your click) or Auto (acts alone inside strict limits)
- Every finding shows its confidence, the factors behind it, the evidence lines, alternatives and what would change its mind
- Guardrails: daily action cap, cooldown, quiet hours, a deny list, and restarts only if you allow them
- Every action is journaled, verified afterwards ("was the problem really fixed?") and can be undone
- Learning statistics: how often Autopilot was right, how often you rejected or undid it, by problem type, with a suggested confidence threshold that is only ever applied by you
- Your own rules: "if this error appears, then notify me" or "then propose this action", with a test button; a rule can never do more than the guardrails allow
- Rejected means quiet: a rejected proposal is not made again until the evidence gets worse or your chosen pause (1 hour to forever) ends
- NeoDash adds fleet warnings: if a plugin causes trouble on one server, it checks every other server that runs the same plugin and warns you in advance
Security That Grows With Your Team
- Two-factor login for every user: authenticator apps (QR code generated locally, no external service), ten one-time recovery codes, and a policy for everyone or for selected roles with a grace period
- Passkeys: sign in with a fingerprint, phone or security key; passkeys can replace password plus code
- Sessions and devices: see where you are signed in, sign out single devices, and get an alert when a new device appears; admins see everyone's sessions
- IP allow-list: restrict the panel to your networks (IPv4 and IPv6), with trusted-proxy support, a "would I lock myself out?" preview and a console switch (
/dash allowlist off) for emergencies
- Signed audit log: every entry is chained to the one before it and signed regularly; export it and verify that nothing was changed, in the panel or offline with the included verifier
- Guided secret rotation: replace the bridge secret between Dash and NeoDash with an automatic connection test and rollback, re-test Discord webhooks, and see which API tokens are due for rotation
- Scoped API tokens: each token carries only the scopes you give it and never more than its owner may do
Community Tools
- Your own public website. Switch on the public site and the status page, whitelist applications, ban appeals, support tickets and player reports are served on a separate port that exposes nothing else: login, panel, API and staff pages answer 404 there. Point your domain at it with nginx, Caddy or Cloudflare, for example . The main admin edits every path, the site title and the public base URL in the panel and changes apply immediately. A dashboard card and copy/open buttons on the staff pages show where the site lives, and
/dash site prints the addresses. On Paper, players who are not whitelisted are told in the kick message where to apply, and banned players where to appeal.
- Restart button: with Dash linked to NeoDash the restart no longer hangs on "restarting" (the browser now opens NeoDash's confirm page), NeoDash's own restart and stop controls send the current server, and Dash without a restart script says honestly that the server will only stop. NeoDash fleet: opening a server interface is faster and no longer needs several clicks. Lag detective: the page no longer reloads every few seconds while a diagnosis runs.
Everything here is off by default and only the Main Admin can switch it on.
- Public status page: real server state, uptime over 24 hours, 7 and 30 days, incidents and planned maintenance, a JSON feed and an embeddable badge; every widget is opt-in
- Whitelist applications: a public form with abuse protection, staff review, approve with one click and a status link for the applicant
- Ban appeals: banned players can appeal without learning anything about other names; staff see the ban details and can unban or reject with a cooldown
- Chat moderation: word filters that see through tricks like leetspeak and spacing, link protection, spam rules, an escalating mute ladder and an overview of suspicious players, with data kept only as long as you choose
- GDPR tool: export everything Dash knows about a player as a ZIP, or erase it with a preview and typed confirmation; the audit chain stays intact and receipts are signed
AI access: A Standard MCP Server for Your Own Assistant
Dash no longer ships its own AI. Instead it speaks the open Model Context Protocol (MCP), so the assistant you already use (Claude, ChatGPT-compatible clients, Cursor, scripts) can look at and, if you allow it, operate your server. It is off by default; only the Main Admin switches it on in AI access.
- Streamable HTTP endpoint
POST /mcp (JSON-RPC 2.0), tools plus read-only resources
- Authentication with the scoped API tokens of API access: a token acts as its owner and never does more than the owner may
- Read tools: status, performance history, console, players, profiles, Guardian lookup, Intelligence diagnosis, Autopilot status, backups, plugins, files (panel data, keys and secrets are never readable), audit log, tickets
- Action tools: console command (with the API deny list and guardrails), kick, ban, unban, whitelist (a reason is mandatory), Autopilot approvals, backups, tickets
- Every tool states whether it is read-only or destructive; every tool can be switched off individually; a kill switch cuts everything at once
- Secrets, tokens and addresses are masked in all output, results are size-capped, calls are rate limited and every call is written to the audit log
- Browser pages are refused unless their origin is allowed (cross-site and DNS-rebinding protection)
- Copy-and-paste connection snippets and the presets "MCP read-only" and "MCP operator" on the page
Deterministic crash and log diagnostics remain available in Maintenance.
A Dashboard That Fits Your Team
The first-run setup provides three sensible starting profiles:
- Light: dashboard, console, players, files, extensions, updates and essential settings
- Normal (recommended): Light plus tickets, notifications, graphs, users, audit and plugin settings
- Full functionality: Normal plus every advanced administration workspace
Intelligence, Maintenance and Guardian can be switched on or off individually through feature availability settings.
The sidebar is customizable directly where it is used:
- Drag items with a mouse or touch and long-press interaction
- Reorder entries to match the team's working sequence
- Hide unused items
- Restore hidden items from the customization list
- Keep the chosen order locally between visits
English, German, Spanish and further interface languages are available through the user language selector.
NeoDash: Optional Multi-Server Management
NeoDash is the central control plane for the Dash ecosystem. Install it when you operate more than one server or want a dedicated Linux-host management surface.
NeoDash repository:
With NeoDash, you can:
- View connected servers and fleet health in one place
- Start, stop, restart, create, install and scan server directories
- Provision Paper, Fabric, Quilt, NeoForge and Vanilla servers
- Select Minecraft 1.21 through 1.21.11, 26.2 or 26.3 with the appropriate Java runtime
- Resolve the correct stable Dash artifact from Modrinth by loader and exact Minecraft version
- Manage fleet users, roles, assignments, audit history, groups and updates
- Run verified backups, scoped restores and optional Google Drive mirroring
- Open Dash, FabricDash and ForgeDash through signed, replay-protected SSO
- Keep plugin dashboard return links on the configured public HTTPS NeoDash domain
NeoDash is optional. A single server can run Dash, FabricDash or ForgeDash perfectly well on its own.
NeoDash quick install
curl -sSL https://raw.githubusercontent.com/Framepersecond/NeoDash/main/install.sh | bash
Native Linux is recommended. Docker artifacts are also available. Java 25 is required when NeoDash directly manages Minecraft 26.2 or 26.3 workloads.
Compatibility and Installation
Choose the build that matches both the server loader and exact Minecraft version.
| Server |
Artifact |
Runtime |
| Paper, Purpur, Spigot, Bukkit 1.21.x |
dash-*.jar |
Java 21 |
| Paper, Purpur, Spigot, Bukkit 26.2 |
dash-*.jar |
Java 25 |
| Paper, Purpur, Spigot, Bukkit 26.3 |
dash-*.jar |
Java 25 |
| Fabric 1.21 through 1.21.11 |
matching fabricdash-mc1.21.x-*.jar |
Java 21 and Fabric Loader 0.19.5+ |
| Fabric 26.2 |
fabricdash-mc26.2-*.jar |
Java 25 and Fabric Loader 0.19.5+ |
| Fabric 26.3 |
fabricdash-mc26.3-*.jar |
Java 25 and Fabric Loader 0.19.5+ |
| NeoForge 1.21 through 1.21.11 |
matching forgedash-mc1.21.x-*.jar |
Java 21 |
| NeoForge 26.2 |
forgedash-mc26.2-*.jar |
Java 25 |
| NeoForge 26.3 |
forgedash-mc26.3-*.jar |
Java 25 |
Paper, Spigot, Purpur and Bukkit
- Download Dash.
- Place the JAR in
plugins/.
- Restart the server.
- Run
/dash register as an operator.
- Open the one-time setup link and create the first administrator.
Fabric and NeoForge
- Download the artifact built for the exact Minecraft version.
- Place it in
mods/.
- Restart the server.
- Run
/dash register as an operator.
- Complete the one-time setup.
Security Notice
Dash is powerful because it can control a real server. Treat the dashboard with the same care as console or host access.
- Use HTTPS or a trusted reverse proxy
- Set
report-url to the public HTTPS dashboard origin before sharing report links
- Use strong, unique credentials and protect the Main Admin account
- Grant the smallest practical role to each operator
- Review the audit log and notification delivery health
- Keep console, files, updates, backups and advanced mutation permissions restricted to trusted staff
- Give assistants the smallest token scopes that do the job (start with "MCP read-only")
- Turn on two-factor login for every account, and keep a copy of the audit log public key off the server
Dash enforces authenticated sessions, CSRF protection, server-side permissions, canonical path checks, bounded requests and uploads, protected-file rules, guarded archive handling, digest-aware downloads and action audit logging. Sensitive key material stays server-side and is masked in the interface.
Partner
Dash is proudly partnered with EmeraldHost, a Minecraft server host. Use the code Frxme10 for 10% off your server.
If Dash saves you time, you can support its development on Buy Me a Coffee. Thank you!
Projects
-
Dash - Paper, Purpur, Spigot, Bukkit and compatible servers
-
FabricDash - Fabric servers
-
ForgeDash - NeoForge servers
-
NeoDash - optional central control plane
Dash gives a single server a serious web administration surface. Add NeoDash when that server becomes a fleet.
评论 0
登录后参与评分
还没有评论,来抢沙发吧