0.9.0 is here, and it brings the most wanted herdr feature:
all your machines running herdr, in a single client 🎉
control your agents and projects across local and remote machines without jumping between terminal tabs.
大佬的总结很及时,最近正好在研究 AI Agent,主要出于以下几点考量:
1. AI 现在写的代码让我不是很满意,一直想找更合适的工具或方法来让 AI 写出更好的代码。遗憾的是,目前并没有让我满意的方案,因此研究 AI Agent 的实现原理就非常有必要,这能让我们在必要时具备为项目做定制的能力。
2. 现在用 AI 做产品,虽然很多时候只是让 AI 做设计写代码,但是仍然有很多业务方向需要把 AI 能力本身集成进产品中,特别是 FDE 正火热的时候,学习 AI Agent 原理也能确保以后能具备为垂直行业定制 AI Agent 的能力。
AI 时代,都说“前端已死”,但是我最近研究 AI Agent 相关技术栈却得出了截然不同结论: AI 时代,前端不仅没死,懂前端的人反而能活的更好!
具体来说,前端已死实际上指的是死掉的前端岗位的旧价值::切页面、写组件、做 CRUD,这些工作正在被 AI 大量接管。
但前端程序员过去积累的 JavaScript/TypeScript、Node.js、浏览器、UI、工程化能力并没有消失,反而直接进入了 AI Agent 的主战场。
今天大量 AI Agent、AI Coding、Agent Harness 和 AI 产品,本身就建立在这套技术生态之上。
这就造成了一个很有意思的差距:不懂前端的人,是在“用 AI 做产品”;懂前端的人,却可以“接管 AI 做出来的产品”。
所以我觉得,AI 消灭的是“前端作为工种”的一部分价值,却放大了“前端作为技术能力”的价值。
以前学前端,是为了找一份前端工作。现在懂前端,却可能成为你进入 AI Agent,甚至自己做 AI 产品的一张非常好的门票。
前端这个岗位也许真的在死,但前端程序员的能力边界,反而被 AI 打开了。
4. 做好供应链管理:这是最容易被忽视的地方。现在 AI 时代,人人都是开发者,都想自己搞开发做产品但是又不得不依赖这几十年打下的繁荣的软件生态。特别是那些用前端技术栈做网站的,你用 AI 开发个网站看起来很轻松,实则一个小小的网站背后依赖了各种各样的第三方包和工具,它们深度参与甚至作为可执行代码的一部分随着你们的网站一起运行。如果这些工具和包有一个被投毒,损失都是灾难性的。
这哥们好惨,一两百万就这么没了!大家天天把 AI 玩的这么溜,有谁真正把安全问题当回事了?
AI 帮你安装的任何软件、SKILL、下载并被读取的任何文档都有可能被意外投毒。
天天使用远程控制软件的,家里给服务开大门的都可能成为黑客攻击的靶子。
为了方便开发关防火墙、关安全中心的行为也都是在为黑客为病毒发邀请函。
你用 AI Vibe Coding 出来的产品上线后大概率也会出现一堆的安全漏洞。
AI 时代给了我们机遇更给了我们危机。
To the hacker who attacked me on Sep 1:
I haven't had any income for 3 years. the $170k you transferred out is all my savings — the money supports me write my code 16+hrs per day for chasing my dream.
I have to move on, I need to provide for my family and keep chasing my dream. I would appreciate it if you could return some of it, for the rest, you won.
for the timeline:
1. 6am UTC, you withdrew my funds from @Morpho
2. 9am UTC, you exchanged them into ETH on @base and into BTC, I had a 3-hour window, but I didn't get my alerts about the activities of my account, that is my fault
3. 9am UTC — same time, you hacked all my twitter accounts.
I still have no clue how I got hacked. over the past few months, 3 apps were installed on my mac: one from a well-known company, one a well designed OSS project and one developed by my previous colleague's kids. I chose to trust all three.
so it's very likely you came in through the supply chain? but even the ACP I was using these days was the official version and pinned. with @claudeai's help, it suggested you may have used osascript + System Events to do the phish for my passwd, then take my @googlechrome sessions.
I was heavily developing my desktop app, that is so sad, but you reminded me that the sandbox env for agents is more important than I thought, which leads me to focus more on this part in @crabtalk_ai , thanks for that point.