Email Server

Product highlights

产品特色

At a glance

一眼看懂

Send & receive email	收发邮件	Standard mail protocols on the usual ports. Any desktop or mobile mail app can connect — you only ever type your own server address.	标准邮件协议、标准端口。任何桌面或手机邮件 App 都能连——你只需要填自己的服务器地址。
Automatic junk blocking	垃圾邮件自动拦截	21 weighted rules score every incoming message. Suspicious mail is flagged, junk is quarantined, and nothing is ever deleted — you can review and restore every message.	21 条加权规则给每封来信打分。可疑的标记出来，垃圾的进隔离区，而且什么都不删——每一封都能查看、恢复。
Your data stays yours	数据只属于你	Mail is stored as individual files on your machine. The built-in filtering looks at sender identity, routing and structure — not at your message body. Reading a body sample only ever happens if you switch on the optional AI pass.	邮件以独立文件形式存在你自己的机器上。内置过滤看的是发件人身份、路由与结构，不看正文；只有当你主动开启可选的 AI 二审时，才会读取正文样本。
No command line needed	无需命令行	A graphical control panel covers status, live logs, quarantine, reputation, blocking and settings. A phone app is available for on-the-go checks.	图形管理面板一站式搞定状态、实时日志、隔离区、信誉、封禁与设置；另有手机 App 可随时随地查看。
Runs as a service	服务方式运行	Installs as a native service: starts with the machine, restarts itself after a crash, no one needs to be logged in.	以系统服务方式运行：开机自启、崩溃自动重启，无需有人登录桌面。
One-time licence	一次性授权	No monthly fee and no per-mailbox charge. Run as many mailboxes as your machine needs.	没有月费，也不按邮箱数量收费。你的机器能装多少邮箱就装多少。

Feature detail

功能详解

1 · Mail transfer — the core

1 · 邮件收发（核心）

• Full protocol set: SMTP for sending, submission with authentication for your own users, POP3 and IMAP for collecting mail. Eight listening ports in total.
• 协议完整：发信走 SMTP，本机用户走带认证的提交端口，收信支持 POP3 与 IMAP，共 8 个监听端口。
• Encryption on every channel: STARTTLS on the standard ports plus implicit TLS ports for clients that prefer an always-encrypted connection.
• 全通道加密：标准端口支持 STARTTLS，另有隐式 TLS 端口给「连接即加密」的客户端使用。
• IMAP keeps devices in sync: read a message on your phone and it shows as read on your desktop — folders and flags included.
• IMAP 多设备同步：手机上读过的信，电脑上也是已读；文件夹与标记状态一并同步。
• POP3 when you prefer downloading: fetch and keep a local copy, with optional leave-on-server behaviour per mailbox.
• POP3 下载模式：把邮件取回本地保存；可按邮箱设置是否在服务器保留副本。
• Multiple domains, many mailboxes: host several of your own domains on one server, each with as many accounts as you need.
• 多域名、多邮箱：一台服务器托管多个自己的域名，每个域名下账户数量不限。
• Outbound delivery with automatic retry: messages go straight from your server to the recipient's server; temporary failures are queued and retried instead of being bounced immediately.
• 外发投递与自动重试：邮件直接由你的服务器投给对方服务器；临时失败会进队列重试，而不是立刻退信。
• Size limits you control: per-message size limit (default 25 MB) with a clear rejection message when exceeded.
• 体积上限可调：单封邮件大小上限默认 25 MB，超限时给出明确的拒收原因。
• One file per message: mail lives as individual files on disk, so backup, restore and migration are just file operations — no database lock-in.
• 一封邮件一个文件：邮件以独立文件落盘，备份、恢复、迁移都只是文件操作，不被数据库绑死。

2 · Junk protection — scored in layers

2 · 垃圾邮件防护（多层评分）

Every incoming message is scored by 21 weighted rules, and the sender's identity is verified before the message is even considered. Two thresholds decide what happens: a low score for tagging, a high score for quarantine.

每封来信都由 21 条加权规则打分，并且在「考虑内容」之前先验证发件人身份。两个阈值决定处理方式：达到低阈值就标记提示，达到高阈值就进隔离区。

3 · Confirmed junk, human-verified

3 · 经人工确认的拦截（可学习）

• Interceptions are not final: you review them, and everything you confirm as junk feeds a separate confirmed-junk store. That store is your evidence base for tuning rules and for telling a legitimate sender apart from a repeat offender.
• 拦截不是终点：你逐件复核，确认是垃圾的进入独立的「已确认垃圾」库。这个库既是调规则的依据，也是区分「正常发件人被误伤」和「惯犯」的证据。
• False positives are released and can be marked as trusted senders, so a real business partner never gets stuck in the filter twice.
• 误拦可以放行，并可标记为可信发件人——真正的合作伙伴不会被同一个过滤器卡第二次。

4 · Optional AI second opinion

4 · 可选的 AI 二审

• For the cases rules cannot judge — complex links, semantics, context — an optional AI pass can give a second opinion on quarantined mail only.
• 规则判断不了的疑难件——复杂链接、语义、上下文——可以由可选的 AI 再给一次意见，而且只审已经被拦下的邮件。
• It can run against a model on your own machine or your own private endpoint, so message context never leaves your network. Off by default; enable it when you want it.
• 可以接你自己机器上的模型或你自己的私有端点，邮件上下文不出你的网络。默认关闭，想用时再开。
• The decision stays human: the AI flags or explains, you confirm.
• 最终决定权仍在人：AI 只提示和解释，由你确认。
• What the AI sees: when you enable this pass, a sample of the message body (up to 4,000 characters) is sent to the model together with the sender, recipient and subject. With a remote endpoint that content leaves your server; with a model on your own machine it stays there. Enabling it is your decision, and the risk is yours to weigh.
• AI 会看到什么：启用后，邮件的正文样本（最多 4000 字符）会连同发件人、收件人、主题一起发送给模型。用远程端点，这些内容会离开你的服务器；用本机模型则留在本机。是否启用完全由你决定，风险也由你权衡。
• Content analysis is never part of the built-in protection: without this pass, the server judges mail purely from envelope, headers and connection data.
• 内容分析永远不是内置防护的一部分：不开这个二审，服务器只依据信封、头部与连接数据判断邮件。

5 · Security and abuse protection

5 · 安全与滥用防护

• Encrypted transport everywhere: all receiving, sending and authentication channels support TLS 1.2/1.3 — passwords and mail content are protected in transit.
• 全程加密传输：收信、发信、认证通道全部支持 TLS 1.2/1.3——密码与邮件内容在传输中受保护。
• Bring your own certificate (a free one works) with the server and key configured in plain files — no vendor lock-in, no hidden certificate service.
• 证书自带（免费证书即可）：证书与私钥就是两个文件，可随时替换——不绑定任何服务商，也没有藏起来的证书服务。
• Brute-force guard: configurable failure count within a time window → the source address is blocked, with the block persisted across restarts and every decision written to the audit log.
• 暴力破解防护：可配置「时间窗内失败几次」→ 自动封锁来源地址；封禁在重启后依然有效，每次判定都写入审计日志。
• Port-level attack blocking: sources that hammer the ports are banned, so scanning never grows into a compromise attempt.
• 端口级攻击拦截：高频冲击端口来源自动封禁，扫描不会演变成入侵尝试。
• Evasion detection: a source rotating through many different identities to keep guessing gets caught and reported as a coordinated attempt.
• 逃逸检测：不断变换身份、换着法子试的来源会被识别出来，并按有组织攻击上报。
• Trusted networks: loopback, your LAN and any address you list are exempt from automated blocking, so you cannot lock yourself out.
• 可信网络豁免：本机、你的局域网以及你列出的地址不参与自动封禁，自己不会把自己关在门外。
• Spoofing resistance: envelope/header sender consistency checks plus signing of your own outbound mail, so your domain is harder to impersonate.
• 防伪造：信封与信头发件人一致性检查，加上对外发邮件签名，别人更难冒充你的域名。
• Salted-hash password storage for every account, with per-account authentication auditing.
• 密码加盐哈希存储，每个账户的认证过程都可审计。

6 · Privacy and data sovereignty

6 · 隐私与数据主权

• Message files stay on your machine. There is no cloud middleman in the path — no third party sees your mail.
• 邮件文件只存在你的机器上，链路上没有云端中转——没有任何第三方看得到你的邮件。
• We do not read your message body. Filtering works on sender identity, routing metadata and structural patterns; reputation tracking stores sender domains, never message content.
• 我们不读你的正文。识别基于发件人身份、路由元数据与结构特征；信誉记录只存发件域名，不碰邮件内容。
• Backups are plain files you can copy yourself and restore anywhere — you are not dependent on us to get your own mail back.
• 备份就是普通文件：你自己就能复制、能恢复到别处——取回自己的邮件不需要依赖我们。

7 · Auditing and observability

7 · 审计与可观测性

• Complete session-segmented log: every connection is logged as one self-contained block — start, connection source, every command issued, end. No interleaving between sessions, so a parser or an auditor can trust the pairing.
• 会话分段完整日志：每个连接是一个自包含的段落——开始、连接来源、每条命令、结束，段与段之间不交叉，解析器和审计人员都能放心配对。
• Structured audit mirror: sessions, authentication attempts, individual messages and anomalies are written to queryable tables alongside the raw log — the raw text stays authoritative and is never filtered or trimmed.
• 结构化审计镜像：会话、认证尝试、单封邮件与异常事件写入可查询的表，与原始日志并存——原始文本始终是权威，不做过滤或裁剪。
• Trace a message backwards: from a delivered message back to its session, its authentication result and the agent that sent it.
• 可回溯：从一封已投递的邮件，追到它所属的会话、认证结果和发信来源。
• Live view: running state, listening ports, send/receive counters, active connections and a real-time log tail in the panel.
• 实时视图：面板里可看运行状态、监听端口、收发计数、活跃连接，以及实时滚动的日志。

8 · Control panel

8 · 图形管理面板

• Overview: status, ports, counters, service health at a glance.
• 总览：状态、端口、计数、服务健康度一眼可见。
• Mail and quarantine: browse, release or delete quarantined messages; manage the confirmed-junk store.
• 邮件与隔离区：浏览、放行、删除被拦邮件；管理已确认垃圾库。
• Security page: see every blocked address with its reverse lookup, unblock with one click.
• 安全页：查看每个被封锁的地址及其反查信息，一键解封。
• Settings and policy: ports, size limits, scoring weights, thresholds, trusted senders and networks — edited in the panel, applied without hand-editing files.
• 设置与策略：端口、体积上限、评分权重、阈值、可信发件人与可信网络——面板里改，不用手工编辑文件。
• Local-only management API: the administrative interface listens on the machine itself; from outside you reach it over an encrypted tunnel instead of exposing it to the internet.
• 管理接口只监听本机：管理面不对公网开放，远程管理走加密隧道，不把后台暴露在互联网上。
• Phone controller app: check status, logs, mail and security — and ask questions — from your phone.
• 手机控制器 App：在手机上查看状态、日志、邮件、安全，并直接提问。
• Built-in assistant: ask the panel about your server in plain language (optionally backed by your own local AI). It answers from diagnostics — it does not read message bodies.
• 内置智能助手：用大白话问它服务器的问题（可接你自己的本地 AI）。它只读诊断数据作答，不读邮件正文。

9 · Deployment and day-to-day operation

9 · 部署与日常运维

• Single-machine deployment: one server, one service, one configuration file you can read.
• 单机部署：一台服务器、一个服务、一份看得懂的配置文件。
• Installs and runs as a native service — auto-start on boot, auto-restart on failure, no logged-in user required.
• 以系统服务方式安装运行：开机自启、故障自动重启，不需要有人登录。
• Archive option: keep a configurable retention window of mail (default 500 days) so history is available without growing forever.
• 归档选项：可配置的邮件保留期（默认 500 天），历史可查又不会无限膨胀。
• Upgrades keep your configuration, accounts and mail in place.
• 升级保留原有配置、账户与邮件。
• What you need: a machine (physical or virtual), a domain you control, the ability to publish DNS records for it, and a provider that allows the standard mail ports.
• 你需要准备：一台机器（实体或虚拟）、一个你自己能管的域名、能修改它的 DNS 记录，以及放行标准邮件端口的网络环境。

Technical specifications

技术规格（给懂行的人）

Protocols	协议	SMTP · Submission (authenticated) · POP3 · IMAP — each with an implicit-TLS variant	SMTP · 提交端口（带认证）· POP3 · IMAP ——各带隐式 TLS 版本
Ports	端口	25 · 587 · 465 · 110 · 578 · 995 · 143 · 993
Encryption	加密	TLS 1.2 / 1.3, STARTTLS and implicit TLS; bring your own certificate and key	TLS 1.2 / 1.3，STARTTLS 与隐式 TLS；证书与私钥自带
Sender verification	发件人验证	SPF / DKIM / DMARC verification, alignment checks, forwarding-aware authentication chains; outbound messages are signed with your own domain key	SPF / DKIM / DMARC 验证、对齐检查、兼容转发的认证链；外发邮件用你自己域名的密钥签名
Anti-spam	反垃圾	21 weighted rules (20 active by default), real-time blocklist lookup, sender reputation, two thresholds (tag at 5, quarantine at 9 — both configurable), quarantine area, confirmed-junk learning store	21 条加权规则（默认启用 20 条）、实时黑名单查询、发件信誉、双阈值（5 分标记、9 分隔离，均可调）、隔离区、已确认垃圾学习库
Abuse protection	滥用防护	Per-IP authentication-failure lockout (window, attempt count and duration configurable, persists across restarts), port-level scan banning, evasion detection, trusted-network exemptions	按 IP 的认证失败锁定（窗口、次数、时长可配，重启后仍生效）、端口级扫描封禁、逃逸检测、可信网络豁免
Authentication	认证	SMTP AUTH (LOGIN / PLAIN), salted-hash password storage, per-account audit trail	SMTP AUTH（LOGIN / PLAIN），密码加盐哈希存储，账户级审计
Storage	存储	One file per message; optional archive with configurable retention (default 500 days); per-message size limit (default 25 MB)	一封邮件一个文件；可选归档，保留期可配（默认 500 天）；单封体积上限默认 25 MB
Audit	审计	Raw session-segmented text log (authoritative, never filtered) plus structured tables for sessions, authentication, messages and anomalies	会话分段的原始文本日志（权威、不过滤）＋ 会话 / 认证 / 邮件 / 异常 四类结构化表
Management	管理	Graphical desktop panel, local-only management API, phone controller app, optional AI assistant	桌面图形面板、仅本机的管理接口、手机控制器 App、可选 AI 助手
Platform	平台	64-bit Windows (desktop or server), runs as a service	64 位 Windows（桌面版或服务器版），以服务方式运行
Licensing	授权	One-time licence, no subscription and no per-mailbox fee — ask us for current terms	一次性授权，无订阅、不按邮箱收费——具体条款请咨询我们

Who it is for

适合谁用

What we do not claim

我们不做夸大

• Automatic certificate issuance is not built in — you supply the certificate (a free one works), and we document the exact steps.
• 不内置自动签发证书——证书由你提供（免费证书即可），步骤我们写清楚。
• AI content checking is optional and off by default; the rules engine is what protects you out of the box. When it is on, the model reads a body sample — we say so plainly rather than hiding it.
• AI 内容判断是可选项且默认关闭；开箱即用的防线是规则引擎。它开着的时候会读取正文样本——这一点我们直说，不掩盖。
• Reputation blocking means some of your outbound mail depends on your domain's DNS reputation — the panel tells you what to publish, but you publish it.
• 信誉拦截意味着你的外发送达率取决于自己域名的 DNS 信誉——面板会告诉你该发布什么记录，但发布由你完成。