How do you make Opus cut the fluff and reply like a human?
热度趋势
趋势数据积累中
百分比基于当前可用热度信号，而非评论数或独立用户人数。
用户认为 Opus 的输出中存在过多的“冗余和噪音”，以及“明显由 AI 生成”的、人类不会使用的措辞。这个问题在 Opus 4.8 中尤为突出，尽管在 Opus 5 中略有改善。社区正在寻求解决方案，例如使用自定义代理或技能来消除噪音，或者利用“caveman”等第三方框架，以使 Opus 的回复更像人类。
Reading through the output of Opus and trying to decipher it, sometimes feels exhausting. There's so much fluff and noise, but the worst is the "painfully AI-generated" words and phrases that it uses, which no human on earth would use:
Furthermore, delving deeper into the implementation reveals that mapping over this collection will ultimately yield a brand new array.
It is crucial to remember that a well-defined repository boundary is a testament to clean architecture.
This was a big issue in Opus 4.8 which seems to be slightly improved in Opus 5.
How do you work around this? Do you use custom agents or skills, asking to remove the noise and fluff? Do you use some kind of 3rd party framework (e.g. caveman)?