This code was completely AI generated utilizing a golang reference implementation json specs to drive correctness and a carefully constructed base prompt and series of steps/prompts
Check out the github in the final link if you want to see those prompts they are checked in in the .claude folder as well as the prompts/ folder
The biggest lesson I learned here is sometimes more context is not better. Providing the LLM specifically with a golang implementation performed much better than providing 6 implementations from 6 different languages
1,835
7
本頁面內容由第三方提供。除非另有說明，OKX 不是所引用文章的作者，也不對此類材料主張任何版權。該內容僅供參考，並不代表 OKX 觀點，不作為任何形式的認可，也不應被視為投資建議或購買或出售數字資產的招攬。在使用生成式人工智能提供摘要或其他信息的情況下，此類人工智能生成的內容可能不準確或不一致。請閱讀鏈接文章，瞭解更多詳情和信息。OKX 不對第三方網站上的內容負責。包含穩定幣、NFTs 等在內的數字資產涉及較高程度的風險，其價值可能會產生較大波動。請根據自身財務狀況，仔細考慮交易或持有數字資產是否適合您。