memory_search每次都调用 LLM 过滤，发送的每个命令都会增加1-2s延时？

[jimaoUncle]

这个能不能优化一下

[CaralHsi]

Hi!

You can use mode = fast instead, so that the search duration would be faster.

{
"query": "...",
"user_id": "...",
"readable_cube_ids": ["..."],
"mode": "fast",
"top_k": 10
}