← 返回首页
关于
关于 Scholardo
Scholardo 是为 agent 研究工作流打造的工作台,支持 macOS 与 Windows。它建立在一个判断之上:AI agent 应当是研究的参与者,而不是文献管理器旁边多出来的一个聊天框。
为什么做这个
今天的研究工具被切成了两半。文献管理器存着你的库、你的标注、你正在读的那一页,却对 agent 一无所知;agent 有能力读、能写、能跑代码,却看不见你的任何一份材料——你只能把段落复制粘贴过去,再把结果粘贴回来。
Scholardo 取消了这次搬运。Agent 团队与研究者共用同一个文献库、同一套标注、同一份实时阅读上下文:通过 40 多个原生 MCP 工具,应用内置的 Agent 团队——或你更习惯的任何兼容 MCP 的 CLI——都能直接检索你的库、读到你划的高亮、建立和更新笔记,并且知道你此刻打开的是哪个文件、停在第几页;对话 Agent 起草时会用引用 key 引用你文献库里的文献。(一条边界说在明处:标注对所有 agent 都只读——它们读你划的高亮,但从不替你划。)这不是把聊天窗嵌进阅读器,而是把研究材料本身开放给 agent。
怎么构建的
- 两个平台,同一个应用:macOS 14 及以上(Apple Silicon 与 Intel)与 Windows 10 及以上(x64),功能完全一致。
- 本地优先:每个项目就是你硬盘上的一个普通文件夹,文献、标注、笔记、会话记录都是里面的文件,检索用的索引也在本机、随时可以重建。没有云同步,本地使用不需要账号。
- 零遥测:应用不做使用统计、行为分析或崩溃上报。细节见隐私政策。
- Key 由你自备:AI 功能用你自己的模型 API Key,存在系统的凭据库里(macOS 的钥匙串、Windows 的凭据管理器),请求直连提供商。检索、标注、引用、网页抓取这些核心功能完全不需要 Key。
靠什么维持
Scholardo 是闭源商业软件,独立开发、自筹资金。收入只有一个来源:软件授权——订阅或买断。没有广告,不出售数据,没有投资人需要交代增长曲线。
这一点与上面的技术选择是同一件事:一个不靠你的数据赚钱的产品,才有条件把数据完全留在你的机器上。
你的材料、你的模型、你的机器。我们只负责把它们接在一起。
产品问题、账单问题、退款、团队授权,都可以发邮件至 support (at) scholardo.com,或使用官网的联系表单。使用方法与教程见文档站。
相关条款:隐私政策 · 服务条款
← Back to home
About
About Scholardo
Scholardo is the workbench for agentic research workflow, on macOS and Windows, built on one bet: an AI agent should be a participant in the research, not a chat window bolted onto a reference manager.
Why it exists
Research tooling today is split down the middle. The reference manager holds your library, your annotations and the page you are on, and knows nothing about your agent. The agent can read, write and run code, and cannot see a single one of your sources — so you copy a passage across, and paste the answer back.
Scholardo removes that errand. The agents and the researcher share one library, one set of annotations, and the same live reading context: through 40+ native MCP tools, Scholardo's agent team — or any MCP-compatible CLI you prefer — can search your library, read the highlights you made, create and update notes, and see which file you have open and which page you are on; when the Conversation Agent drafts, it cites works from your library by their citation keys. One limit, stated plainly: annotations are read-only to every agent — they read the highlights you made but never make them. This is not a chat window embedded in a reader — it is the research material itself made available to the agent.
How it is built
- One app, two platforms: macOS 14 and later (Apple Silicon and Intel) and Windows 10 and later (x64), with the same features on both.
- Local first: each project is an ordinary folder on your own disk, and sources, annotations, notes and session history are files inside it; the search index is local too, and can be rebuilt at any time. No cloud sync, and no account for local use.
- No telemetry: the app collects no usage statistics, behavioural analytics or crash reports. The privacy policy has the detail.
- Your keys, your calls: AI features run on API keys you supply, held in the system credential store (the Keychain on macOS, Credential Manager on Windows), with requests going straight to the provider. Search, annotation, citation and web capture need no key at all.
How it is funded
Scholardo is closed-source commercial software, independently built and self-funded. It has exactly one revenue source: software licences, by subscription or one-time. No advertising, no data sold, and no investors to show a growth curve to.
That is the same decision as the ones above, seen from the other side: a product that does not make money from your data can afford to leave all of it on your machine.
Your material, your model, your machine. We only wire them together.
Product questions, billing, refunds and team licensing all reach a human at support (at) scholardo.com, or through the contact form. For how to use the app, see the documentation.
Also here: Privacy Policy · Terms of Service