Most of chat requests get abrupt closer to the end

Where does the bug appear (feature/product)?

Cursor CLI

Describe the Bug

Most of the time when I ask LLM for something and it responds, it does not finish its response, it gets stopped quite often in the middle of editing a file. Sometimes it’s gonna get stopped after it’s finished everything, but it all depends on a chance. Mostly it happens with Claude. It did not happen even once with Sonic when it was available.

Steps to Reproduce

Open your project. In my case it was TypeScript project. Open a chat. Write a prompt. Wait for response. The response will start streaming, but at some point it will just stop for no reason.

Expected Behavior

I expected it to finish streaming reasonably with a logical finish.

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 1.5.7
VSCode Version: 1.99.3
Commit: 6aa7b3af0d578b9a3aa3ab443571e1a51ebb4e80
Date: 2025-08-27T22:43:32.248Z (1 day ago)
Electron: 34.5.8
Chromium: 132.0.6834.210
Node.js: 20.19.1
V8: 13.2.152.41-electron.0
OS: Darwin arm64 23.6.0

Additional Information

I still can use it (I just ask it “please go on” every time), and it’s still way better than to work without Cursor (when you have to copy and paste a lot to chat GPT). but it’s not as smooth and comfortable as it was even a week ago or two…

Does this stop you from using Cursor

No - Cursor works, but with this issue

Hey, sorry to hear you’ve been facing this!

Can you confirm if this is specific to a certain model, or if this happens with many?