Lectern 隐私政策
Lectern 是一个辅助阅读的浏览器扩展：记录你读文章的进度和专注时间，划词翻译并讲解，把查过的词做成卡片供复习。开发者不运营任何服务器，也收不到你的任何数据。
保存在你设备上的数据
下面这些保存在浏览器本机的存储里，不会自动离开你的设备：
- 被判定为文章的网页：网址、标题、正文，以及你在上面的阅读进度和专注时间。
- 你划词查过的内容、模型给出的翻译和讲解、由此生成的复习卡片和复习记录。
- 你的设置，包括模型服务的 API Key。API Key 不会进入网页，也不会出现在导出文件里。
- 诊断日志（最近的出错现场和翻译耗时）和首页按钮的点击次数。
发送给模型服务的数据
Lectern 用你自己填写的 API Key，直接请求你在设置页选择的那家模型服务（例如 OpenAI、Anthropic、MiniMax、DeepSeek，或你自己填的地址）；没有预设的服务商，中间也不经过开发者的服务器。设置页顶部列着下面这些内容，在你点「同意并开始」之前不会发送任何东西；没有填 API Key 时同样不发送。同意可以随时在设置页撤回。会发送的有：
- 文章判别（自动）：你打开一个网页时，Lectern 会把它的网址、标题和页面文字（最多约 3 万字）发给模型，判断它是不是一篇文章。这一步是自动的。把网站加入设置页的「文章记录黑名单」后，那些网站的页面不会被发送。
- 划词翻译（你选中文字时）：选中的文字、它所在的段落和文章标题。只在翻译白名单里的网站，或你手动为当前页开启后才会生效。
- 截图翻译（你主动框选时）：文字识别在本机完成，截图不上传；识别出的文字会发给模型。
- 用一句话改名单（你点发送时）：你输入的话、最近几句对话、两份名单的内容和当前页的网址。
- 其余用到模型的功能（你主动使用时）：讲解生词、就着一段追问、生成文章回顾、用模型复查或筛选已有记录、请模型建议黑名单。发送的是这次操作涉及的文字：相关的段落或文章正文、标题和网址。
模型服务如何处理这些数据，由该服务商自己的隐私政策决定。
发送给同步服务器的数据（可选）
设备同步默认关闭。只有你在设置页填写了自己部署的服务器地址和 Token，Lectern 才会把上面「保存在你设备上的数据」（API Key 除外）发到那台服务器，用来在你的设备之间合并。这台服务器由你自己或你信任的人运营，开发者接触不到。
我们不做的事
- 不出售、不转让你的数据给任何第三方。
- 不把数据用于广告、信用评估或与辅助阅读无关的用途。
- 不内置任何统计分析或追踪服务。
删除数据
设置页的「清除此设备记录」会删除本机的阅读记录和卡片；卸载扩展会删除它在浏览器里保存的全部内容。同步服务器上的数据要在那台服务器上删除。
联系
有疑问请在 GitHub 仓库提 issue。政策有变更时会更新本页的生效日期。
Lectern Privacy Policy
Lectern is a browser extension that assists reading: it tracks your progress and focus time on articles, translates and explains text you select, and turns looked-up words into review cards. The developer operates no server and receives none of your data.
Data stored on your device
The following is kept in the browser's local storage and does not leave your device on its own:
- Web pages identified as articles: URL, title, body text, and your reading progress and focus time on them.
- Text you looked up, the model's translations and explanations, and the review cards and review history built from them.
- Your settings, including the API key for the model service. The key is never exposed to web pages and is never included in exported files.
- Diagnostic logs (recent errors and translation timings) and click counts for home-page buttons.
Data sent to the model service
Lectern uses the API key you provide to call, directly, the model service you choose in settings (for example OpenAI, Anthropic, MiniMax, DeepSeek, or an address you enter yourself). There is no preset provider, and requests do not pass through any server run by the developer. The settings page lists the items below, and nothing is sent until you click "Agree and start" there; nothing is sent without an API key either. You can withdraw consent in settings at any time. What is sent:
- Article detection (automatic): when you open a web page, Lectern sends its URL, title and page text (up to about 30,000 characters) to the model to decide whether it is an article. This happens automatically. Pages on sites you add to the "article blocklist" in settings are not sent.
- Selection translation (when you select text): the selected text, its surrounding paragraph and the article title. Active only on sites in your translation allowlist, or after you enable it for the current page.
- Screenshot translation (when you draw a region): text recognition runs on your device and the screenshot is not uploaded; the recognized text is sent to the model.
- Editing lists in plain language (when you press send): your message, the last few messages, both lists, and the current page URL.
- Other model features (when you use them): explaining words, follow-up questions about a passage, generating an article review, re-checking or filtering saved records with the model, and asking the model to suggest blocklist rules. What is sent is the text involved in that action: the relevant passage or article text, its title and URL.
How the model service handles this data is governed by that provider's own privacy policy.
Data sent to a sync server (optional)
Device sync is off by default. Only after you enter the address and token of a server you deploy yourself does Lectern send the data listed under "Data stored on your device" (except the API key) to that server, to merge it across your devices. That server is run by you or someone you trust; the developer has no access to it.
What we do not do
- We do not sell or transfer your data to third parties.
- We do not use data for advertising, creditworthiness, or any purpose unrelated to assisting reading.
- We include no analytics or tracking service.
Deleting data
"Clear records on this device" in settings deletes local reading records and cards; uninstalling the extension deletes everything it stored in the browser. Data on a sync server must be deleted on that server.
Contact
Questions: open an issue in the GitHub repository. The effective date on this page is updated when the policy changes.