dex 回应删 wiki 论：文档摘要只在提速不降质时才有必要
I cannot cosign the $32m paper but yes I am nodding along to this one pretty hard
dex 表示无法为“3200 万美元论文”背书，但认同“面向人类的代码库”与“面向智能体的代码库”这一区分框架。他认为文档或按目录层级压缩的摘要只有在提速且不降质时才有必要，否则更简单的做法是接受一点速度损失、每次都从源头读取。他还指出，若代码库闭源，mintlify 等文档可作为用户智能体集成时的“一手来源”。
查看原帖不。我们必须停止把旧的人类中心模式套用到新时代的工作方式上。 删掉你所有的 wiki。 让智能体直接从源头回答问题。 在初始状态下，我们拥有： - 零个仪表盘工具 - 零份文档 - 我们不制作也不使用摘要 我们所做的一切都是原始资料 + 智能 没有会过时的东西 没有需要整理的东西 没有需要导航的东西 始终完全保持最新 完美个性化 答案随着每一个模型而变得更好 如果我们想看图表，我们就提出要求，一分钟内就能拿到 如果我们想要某个问题的答案，我们就提问，然后得到一个精确到分钟的正确答案
查看原文
No. We've got to stop applying old human-centric patterns to working in the new era. delete all your wikis. have an agent answer questions from source instead. at spawn we have: - zero dashboard tools - zero docs - we do not make or use summaries everything we do is primary sources + intelligence Nothing to go stale Nothing to organize Nothing to navigate Always exactly up to date Perfectly personalized The answers get better with every model If we want to see a chart, we ask for it, and we have it in a minute If we want an answer to a question, we ask it, and we have a to-the-minute correct answer
正文
I cannot cosign the $32m paper but yes I am nodding along to this one pretty hard
Only reason for any kind of doc/summary (eg compacted summary per dir down the hierarchy) is if it increases speed without reducing quality , but unless you’re buildings evals specific to your codebase watching like a hawk, simpler answer is live with a little speed hit and just go from source every time
I still can’t get over the overall framing of “codebases for humans” vs “codebases for agents”
And also mintlify/docs in general probably serve as one of those “primary sources” if your codebase is closed and your users’ agents need a primary source to integrate it