Claude 5 模型的上下文工程新规则
[
[
We removed ~80% of the Claude Code system prompt for our newest models, this is what we've learned about writing system prompts, skills and Claude.MDs for them.
我们为最新模型移除了约 80% 的 Claude Code system prompt，以下是我们在为它们编写 system prompt、skills 和 Claude.MDs 方面学到的经验。
The new rules of context engineering for Claude 5 models
Claude 5 模型的 context engineering 新规则
I’ve written previously about how to best prompt the newest generation of Claude 5 models and work with them iteratively to discover what you want to build.
我之前写过关于如何最好地提示最新一代的 Claude 5 模型，并与它们迭代合作以发现你想要构建什么。
But when you send a message to Claude, the prompt is only a small part of the context it gets. Much of your context is assembled from your system prompt, Skills, CLAUDE.md files, memory, and other sources. We call this context engineering, and it makes a big impact on the results you generate when using Claude Code or in building your own agents.
但是，当你向 Claude 发送消息时，提示词只是它获取的上下文的一小部分。你的大部分上下文是由你的系统提示词、Skills、CLAUDE.md 文件、记忆和其他来源组装而成的。我们称之为上下文工程，它对你在使用 Claude Code 或构建自己的智能体时生成的结果有着重大影响。
Unlike a prompt, context is used generally across many requests, so it cannot be as specific. How do you build these general prompts and guidance for Claude, especially when you don’t know what a user’s prompt might be?
与提示词不同，上下文通常用于许多请求中，因此不能那么具体。你如何为 Claude 构建这些通用的提示词和指导，尤其是当你不知道用户的提示词可能是什么的时候？
This can be surprisingly difficult as Claude’s own capabilities evolve. Most recently, we noticed a large jump in the way we prompt the newest generation of Claude models. We removed over 80% of Claude Code’s system prompt for models like Claude Opus 5 and Claude Fable 5 with no measurable loss on our coding evaluations.
随着 Claude 自身能力的不断演进，这可能会出人意料地困难。最近，我们注意到在提示最新一代 Claude 模型的方式上有了巨大的飞跃。对于像 Claude Opus 5 和 Claude Fable 5 这样的模型，我们移除了 Claude Code 超过 80% 的 system prompt，而在我们的代码评估中没有任何可衡量的损失。
Here’s what we’ve learned about prompting this new class of models, and how you can utilize it to update your context engineering. We’ve ...