
Boris Cherny: We Cut 80% of Claude Code’s Prompt
当AI能力狂飙,产品该如何重新开始
54 分钟阅读AI
【AI前沿】删掉80%的提示词之后,Claude Code如何释放模型真正的能力? Claude Code创作者Boris Cherny在Startup School 2026分享了Opus 5带来的能力跃迁,以及AI产品开发正在发生的根本变化。从提示注入防护、系统提示词删减,到长期自主执行、动态调度数千个智能体,他强调:新模型发布后,开发者应敢于清空旧有设计,先观察模型能做什么,再用最少的约束补足问题。文章也将带你理解“产品过载”、验证机制和未来程序员需要培养的产品与协作能力。 🎙️ 主讲人介绍 Boris Cherny:Claude Code创作者,围绕Opus 5、系统提示词、智能体协作与AI产品开发方法,分享一线实践与判断。 Diana Hu:本期对谈主持人,负责采访Boris Cherny,并围绕模型能力、产品设计和开发趋势推进讨论。 核心议题:Opus 5的能力跃迁、系统提示词精简、提示注入与安全、长期自主执行任务、动态工作流与智能体协作、AI时代的开发者能力。 关键词:Claude Code、Opus 5、提示工程、智能体、产品过载、编程教育
章节 1:Opus 5的突破与持续运行能力
📝 本节摘要:
Boris介绍了Opus 5相较前代的新能力:模型经过训练后常会涌现出未被直接教授的技能,其中最突出的是能够结合自动模式长时间持续执行任务,甚至运行数天、数周或数月而不停下来。此外,他还提到模型似乎已不再容易受到提示注入攻击。
[原文] [Diana Hu]: All right, Boris. We're so excited to have you here, the creator of Claude Code.
[译文] [Diana Hu]: 好的,Boris。非常高兴邀请到你——Claude Code 的创作者。
[原文] [Boris Cherny]: Thank you.
[译文] [Boris Cherny]: 谢谢。
[原文] [Boris Cherny]: It's great to be here.
[译文] [Boris Cherny]: 很高兴来到这里。
[原文] [Diana Hu]: Fresh off the press. You guys just shipped Opus 5 yesterday.
[译文] [Diana Hu]: 这可是刚刚发布的消息。你们昨天才推出了 Opus 5。
[原文] [Boris Cherny]: Yes.
[译文] [Boris Cherny]: 是的。
[原文] [Diana Hu]: And it seems that model performance keeps accelerating. You guys got and took ARC AGI 3 to 30%, which is incredible.
[译文] [Diana Hu]: 而且模型性能似乎还在不断加速提升。你们把 ARC AGI 3 的成绩提升到了 30%,这太不可思议了。
[原文] [Boris Cherny]: Yes.
[译文] [Boris Cherny]: 是的。
[原文] [Diana Hu]: And for context, before the best score was in the low single digits or low teens, right? What can Opus 5 do now that it couldn't versus the previous version?
[译文] [Diana Hu]: 作为背景说明,之前最好的成绩还只有个位数或十几,对吧?那么相比上一版,Opus 5 现在能做到哪些过去做不到的事情?
[原文] [Boris Cherny]: Yeah, there's a lot that goes into every new model, and there's a lot of new capabilities that we teach and get the model to do. Whenever you do model training, you try to teach a whole bunch of different things, and most often it doesn't work. But some subset of the things the model does learn, and sometimes it also surprises you. It has these skills, it has abilities that you actually didn't really teach it, but it just kind of learned. For Opus 5, one example of something it does that I think no other model has done is it runs for a very long period of time, and especially when you combine Opus 5 with auto mode, it's just incredible. It can go for days, weeks, months at a time. It just won't stop.
[译文] [Boris Cherny]: 是的,每个新模型都要加入很多东西,我们也会教它、让它掌握许多新能力。训练模型时,你会尝试教它各种各样的事情,但大多数时候并不会成功。不过,模型确实会学会其中一部分,有时还会给你带来惊喜。它会拥有一些你其实没有真正教过它的技能和能力,只是自己慢慢学会了。以 Opus 5 为例,我认为其他模型都没有做到的一点是,它可以运行非常长的时间。尤其是把 Opus 5 和自动模式结合起来,效果简直不可思议。它可以连续运行几天、几周,甚至几个月,完全不会停下来。
[原文] [Boris Cherny]: You don't even need to use scaffolding. So you don't need the slash goal, you don't need all this other stuff. It'll just go because it knows it needs to do the task. Another thing that I'm really excited about, and I'm going to start, I think, to talk about a little bit more, but it's kind of surprising because it's such a new capability, is the model does not seem to be prompt-injectable anymore.
[译文] [Boris Cherny]: 你甚至不需要使用脚手架。所以你不需要输入斜杠 goal,也不需要其他那些东西。它会自己继续执行,因为它知道自己需要完成任务。另一个让我非常兴奋、我想之后开始多谈一点的能力,是模型似乎已经不再容易受到提示注入攻击了。这很令人意外,因为这是一项非常新的能力。
- 章节 2:提示注入与系统提示词精简
- 章节 3:删掉提示词,重新认识模型
- 章节 4:产品过载与解锁模型能力
- 章节 5:给模型更难的任务
- 章节 6:提示工程正在改变
- 章节 7:让Claude自主完成长期任务
- 章节 8:动态工作流驱动数千智能体协作
- 章节 9:编码几乎已经被解决
- 章节 10:计算机专业学生仍需掌握什么
正在加载后续章节…







