Cursor keeps interrupting its responses and can't execute completely at one time

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When I send a request, Cursor disconnects after running for a while. It can’t give me a complete response like it used to.
I can only tell it “Continue, continue, continue” again and again after it interrupts, in order to finally get a complete response.

Steps to Reproduce

It’s easy to reproduce. All you need to do is have a conversation with it.

Expected Behavior

Under normal circumstances, it should provide a complete answer to me or finish writing the code.

Operating System

Windows 10/11

Version Information

Version: 3.5.17 (user setup)
VSCode Version: 1.105.1
Commit: d5b2fc092e16007956c9e5047f76097b9e626ca0
Date: 2026-05-20T02:43:31.559Z
Layout: editor
Build Type: Stable
Release Track: Default
Electron: 39.8.1
Chromium: 142.0.7444.265
Node.js: 22.22.1
V8: 14.2.231.22-electron.0
OS: Windows_NT x64 10.0.19045

For AI issues: which model did you use?

sonnet4.6、compose2.5

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

Hey, this looks like a known bug with a false “previous response was interrupted” message. The agent thinks it was interrupted even though the reply actually finished, then asks you to continue. See a similar thread: Cursor automatically injects "Your previous response was interrupted" after every response

To confirm it’s the same issue, could you please send the Request ID for one of those “interrupted” replies. You can get it like this: right click the chat message > Copy Request ID.

Also, can you confirm:

  • Does this happen on any task, or only on long or complex ones?
  • Which mode is it in, Agent, Ask, or Plan?

We’re tracking this issue on our side, but we don’t have an exact ETA for a fix yet. I’ll post an update in the thread once we have one.

The problems are different. Its claim is that the response is complete. But my problem is that it interrupts on its own before the response is truly complete. And after the interruption, there is no message like “previous response was interrupted”.

Both the plan and the Agent will experience this issue.

Got it, thanks for clarifying. This is a different case. To dig in, could you please send:

  1. The Request ID from one of the interrupted replies right click the chat message > Copy Request ID.
  2. A screenshot from the moment it stops. I want to see if Cursor shows the complete status checkmark, or if it just hangs with no clear finish.
  3. Roughly what kind of task it is long refactor, generating a big file, scanning lots of files and after how many steps or tokens it usually cuts off.

A similar report is here: Cursor doesn't continously execution. If the symptoms match, let me know. With the Request ID, we can more accurately check what happened on the inference side.