Chat gets stuck in an infinite loop responding to itself

Support Bug Reports
, ,
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

While in auto mode, my chat kept getting stuck in an infinite loop, talking to itself. I tried reprompting multiple times. Each time, I had to manually stop it from continuously responding to itself.

Steps to Reproduce

  1. Start a chat in Auto mode Agent
  2. Continue prompting
  3. Observe eventually the chat begins to loop responding to itself.

Operating System

MacOS

Version Information

Version: 2.4.21 (Universal)
VSCode Version: 1.105.1
Commit: dc8361355d709f306d5159635a677a571b277bc0
Date: 2026-01-22T16:57:59.675Z
Build Type: Stable
Release Track: Nightly
Electron: 39.2.7
Chromium: 142.0.7444.235
Node.js: 22.21.1
V8: 14.2.231.21-electron.0
OS: Darwin arm64 24.1.0

For AI issues: which model did you use?

Auto

For AI issues: add Request ID with privacy disabled

4439a15f-7fb7-4a9a-9e8b-0634ee5fd72e

Additional Information

Chat Exerpt:

The triple quotes implementation is good practice.
Yes.
End.
I think that's the end.
The user asked about the practice and I answered.
Yes.
Perfect.
End of response.
I think that's it.
The conversation can move on.
Yes.
End.
Actually, the response already asks what to do next, so the user can respond.
Yes.
Perfect.
The answer is comprehensive.
I think that's it.
The conversation is complete.
Yes.
Done.
Final.
The answer is that using triple quotes is excellent practice.
Yes.
End.
I think the user will understand that their code is good.
Yes.
End.
The response explains why and gives examples.
Yes.
Done.
Final answer.
The triple quotes with chained logger calls is excellent practice.
Yes.
End.
I think that's the end.
The user asked about the practice and I answered

Does this stop you from using Cursor

No - Cursor works, but with this issue

3

Hey, thanks for the report.

Looks like a rare bug where Auto mode loses track of what it already replied, then starts “thinking out loud” in an infinite loop. I flagged this to the team and shared your Request ID for investigation.

A couple questions for context:

  • Does this happen with specific types of prompts, or randomly after N messages?
  • About how many messages are usually in the chat before it starts?

Let me know if you notice any pattern that triggers it.