Currently, the backend logs token consumption for each conversation by splitting it into multiple records. However, since all records share the same type, this segmentation lacks apparent utility and obscures the total token usage per conversation, making it difficult for users to track consumption accurately.
To improve transparency, when a user clicks on a specific conversation’s token record, a dedicated interface could be displayed to present detailed token consumption information for that conversation. This approach would provide clearer insights compared to the current method of fragmented and aggregated logging, which may appear confusing and offers limited practical value.
Not a fan of this idea. Because there is also capability of changing models and re-running the same prompt in a conversation. Also you are able to edit your previous prompt which is kind of a difficult feature to implement.
Conversations are not stored on cursor server from what I know (please correct me if I’m wrong), its stored on your machine. Therefore cursor only has track of the requests you made, not the conversations you had.
Ofcourse, I’m not a Cursor dev, but these are my 2 cents on the matter.
You may not have understood what I mean. I think the current new way of presenting records is too fragmented, making it difficult for people to conduct effective analysis and judgment. It would be best to restore the previous method of one record per conversation.
