Every third output gets cut off!

Bug Reports
1

I understand that Cursor reduces context and output tokens to save costs, but every third output from Claude models gets cut off prematurely for me! Even when I use “Max,” the tool reads files, responds with 2-3 sentences or asks a question, and then stops. It doesn’t start working with the files. I’m not willing to pay for premium tool calls only to receive a summary of the files.

I use detailed rules for response design and how to handle the codebase, but everything is simply ignored.

Has anyone had similar experiences and tips for a solution?

1 Like
2

Hey, can you share the request ID? First, disable private mode, reproduce the issue again, and then share the ID.

3

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.