Show HN: Add semantic caching to LLM APIs with one-line-of-code | Not Hacker News!