Hello Cursor Team,
I truly appreciate the innovative integration of AI Agents and functionalities in the Cursor code editor. However, I’d like to offer some suggestions to improve transparency, which I believe would enhance the overall user experience.
Token Management Transparency
Currently, it’s unclear how many tokens are consumed by the system prompt. While the documentation mentions the context window size, it’s ambiguous whether the prompt tokens are factored into that total. Moreover, the interface does not display how many tokens have already been used during an ongoing session, leaving users unsure of how many tokens remain.
Visibility of Search Results
The platform’s search tools—both the web search and the documentation search—lack transparency. It isn’t evident what results these tools return, which makes it hard to evaluate their effectiveness. For example:
- The web search results are hidden.
- The documentation search only shows a partial result of the parsed content.
- Similarly, the code base search does not clearly reveal what it finds during a query.
Context Window Handling
Another area of concern is how the chat handles situations where the conversation exceeds the context window. When a session has grown long, it’s unclear which parts of the conversation are taken into account—whether it’s the first messages, the latest messages, or perhaps a summarized version of earlier content. Clear insight into how the system manages and processes extended chats would be highly beneficial.
Greater visibility into these areas would help users understand the inner workings of the editor better and provide more informed feedback. I believe that embracing transparency could significantly enhance the product’s usability and foster a stronger connection with the user community.
Thank you for considering these suggestions. I look forward to seeing further improvements in Cursor.