📄 中文摘要
缓存是提高 API 性能和降低成本的重要机制。每当 Claude Code 代表用户发送消息时，都会进行一次 API 调用，这个调用包含了 Claude 响应所需的所有信息。理解缓存的工作原理，尤其是缓存命中和未命中的概念，有助于优化 API 调用的效率。文章深入探讨了 API 调用的结构、缓存的生命周期及其 TTL 定时器，以及如何通过合理的工作结构来提升缓存效果。此外，还指出了常见的缓存反模式，帮助开发者避免在实现过程中可能遇到的问题。
掌握 Claude 代码中的缓存命中
出处: Mastering Cache Hits in Claude Code
发布: 2026年2月21日
📄 English Summary
Mastering Cache Hits in Claude Code
Caching is a crucial mechanism for enhancing API performance and reducing costs. Each time Claude Code sends a message on behalf of the user, it makes an API call that includes all necessary information for Claude to respond. Understanding how caching works, particularly the concepts of cache hits and misses, can significantly optimize API call efficiency. The article delves into the anatomy of an API call, the lifecycle of caches, and the TTL timer, as well as how to structure work for better caching outcomes. Additionally, it highlights common caching anti-patterns to help developers avoid potential pitfalls during implementation.