— 关于本站
关于本站.
写给已经付费、却只用到皮毛的人。
你可能同时订着 Claude、ChatGPT、Cursor、Warp、Kimi, 但每天打开的还是同一个聊天框。Aklman · Library 做的事很窄: 把一份订阅里值得学的入口、可以跳过的长尾、需要换工具的边界, 一本一本讲清楚。
You may be subscribed to Claude, ChatGPT, Cursor, Warp, or Kimi, yet still open the same chat box every day. Aklman · Library has a narrow job: show which parts of a subscription are worth learning, which long tails can be skipped, and where another tool is the better entrance.
— I
为什么存在.
AI 产品页会告诉你它能做什么, 但很少告诉你这件功能值不值得花十分钟学。 一份订阅打开的入口越来越多: 聊天框、桌面端、CLI、IDE、API、agent、 图片、视频、云端项目。问题不再是「有没有功能」, 而是「这件活该从哪里开始」。
Product pages tell you what an AI tool can do, but rarely tell you whether a feature is worth ten minutes of your attention. One subscription now opens many entrances: chat, desktop app, CLI, IDE, API, agents, images, video, cloud projects. The problem is no longer whether a feature exists. The problem is where a piece of work should begin.
这里的每一本书只解决一类判断: Anthropic 会员里哪几面值得打开, OpenAI 一份订阅该怎么分流, Cursor 和 Warp 什么时候像元工具, agent 平台该交给它什么、又该把权限收在哪里。读完以后, 你应该能去做一件事, 或者放心不做一件事。
Each book here resolves one kind of judgment: which surfaces of an Anthropic membership are worth opening, how an OpenAI subscription should be split, when Cursor or Warp becomes a meta-tool, what an agent platform should handle, and where its permissions should stop. After reading, you should be able to do one thing, or confidently not do one thing.
选择短书这种形态, 是为了给判断留边界。博客容易把未想完的东西一段段发出去; 书必须先定目录, 再落章节, 最后收束成一句能站住的话。这里宁愿慢一点, 也不把官网、教程、清单和广告混成同一种东西。
The short-book format gives each judgment an edge. Posts can leak unfinished thoughts one paragraph at a time; a book has to set the outline, land the chapters, and close with a sentence that can stand. This library would rather be slower than mix vendor copy, tutorials, lists, and ads into the same thing.
— II
这里写什么.
第一类是订阅包。Claude、ChatGPT、Google One、Kimi、豆包、通义, 每个产品都在把自己扩成一个入口集合。书里不列完整功能表, 只回答: 最低值得开的方案是什么, 哪个入口该成为默认, 哪些功能看起来热闹但可以跳过。
The first group is subscription bundles. Claude, ChatGPT, Google One, Kimi, Doubao, and Tongyi are all expanding into sets of entrances. The books do not list every feature. They answer which plan is worth opening, which entrance should become default, and which busy-looking features can be skipped.
第二类是工作流。知识、开发、自媒体、学习这些事, 工具只是载体。 好的工作流不是「我个人怎么用」, 而是一把尺: 什么材料该进来, 哪些步骤可以省掉, 哪些自动化会省时间, 哪些自动化只是在制造新的维护成本。
The second group is workflows. Knowledge work, development, publishing, and learning use tools as carriers. A useful workflow is not a personal setup tour; it is a ruler: what should enter, which steps can disappear, which automations save time, and which ones create new maintenance cost.
第三类是 agent 平台和常识手册。OpenCode、Hermes、OpenClaw、Agent 实战, 讲的是把模型接到工具、文件、权限和长期任务上。字面意思这类书, 则把 token、向量、加密、幻觉这些常见词讲到能转述给别人。
The third group is agent platforms and primers. OpenCode, Hermes, OpenClaw, and the agent book are about connecting models to tools, files, permissions, and long-running tasks. Primer-style books make words like token, vector, encryption, and hallucination clear enough to explain to someone else.
— III
怎么读.
— IV
作者与边界.
这个文库由 Aklman 维护。作者写代码, 也写字, 长期关心的是工具怎样进入真实工作, 而不是发布当天的漂亮话。这里默认站在付费用户一侧: 可以推荐, 但要说清为什么; 可以喜欢一个产品, 也必须写它的 trade-off。
The library is maintained by Aklman, a writer and engineer more interested in how tools enter real work than in launch-day polish. The default side is the paying user: recommendations need reasons; affection for a product still has to include its trade-offs.
这里不做 affiliate 排序, 不夹带课程软广, 不为了新模型发布重写整本书。 大改版会补边注, 判断变了会开新章或新书。读到任何一本想说点什么, 章节和 About 页末尾都有评论区, 也可以直接发邮件。
There are no affiliate rankings, no course tie-ins, and no full rewrites just because a new model ships. Major redesigns get notes; changed judgments become new chapters or new books. If you have something to say, comments live at the end of chapters and this About page, or you can email directly.
工具替你做事, 替不了你判断.
Tools can do the work, not the judgment.
Aklman · 2026
— 讨论
讨论.
评论区初始化中…