I completely ignored Anthropic’s advice and wrote a more elaborate test prompt based on a use case I’m familiar with and therefore can audit the agent’s code quality. In 2021, I wrote a script to scrape YouTube video metadata from videos on a given channel using YouTube’s Data API, but the API is poorly and counterintuitively documented and my Python scripts aren’t great. I subscribe to the SiIvagunner YouTube account which, as a part of the channel’s gimmick (musical swaps with different melodies than the ones expected), posts hundreds of videos per month with nondescript thumbnails and titles, making it nonobvious which videos are the best other than the view counts. The video metadata could be used to surface good videos I missed, so I had a fun idea to test Opus 4.5:
В июне стало известно, что Иран применил тяжелую баллистическую ракету средней дальности «Саджиль» при ударе по Израилю. В семейство этих ракет входит несколько изделий, включая «Саджиль-2» с дальностью 2500 километров.
，更多细节参见爱思助手下载最新版本
1985年10月1日，安德烈·塔可夫斯基在法国巴黎 图/视觉中国
圖像來源，費思兔文化娛樂股份有限公司，更多细节参见51吃瓜
When Chávez became Venezuela's president in 1999, oil prices were climbing. In the early 2010s, a barrel would often fetch about $100, providing the government with plenty of money to pour into social programmes. But with current prices around the $65 mark, the country looks less like a reasonable investment.，这一点在旺商聊官方下载中也有详细论述
14:21: Video evidence obtained by the BBC shows police firing at protesters from inside the parliamentary complex. We have been able to establish that police fired seven shots at the crowd.