Auto mode OFTEN "TAKING LONGER THAN EXPECTED"

Support Bug Reports
,
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When i am using auto mode, it often show me “taking longer time than expexted”. After sending it a new dialogue “continue”, it will continue. But sometime the new “continue” will still be same as above.
When using auto mode, it often fall in “reconnecting” too. When context is very long, which capture more than 70% of capacity, this situation will always happen.

Steps to Reproduce

Just regularly use it but in Auto mode.

Expected Behavior

It should response fluently.

Operating System

Linux

Version Information

Version: 2.6.20
VSCode Version: 1.105.1
Commit: b29eb4ee5f9f6d1cb2afbc09070198d3ea6ad760
Date: 2026-03-17T01:50:02.404Z
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: Linux x64 6.8.0-79-generic

For AI issues: which model did you use?

Auto.

For AI issues: add Request ID with privacy disabled

d225f060-c3a8-47e5-b139-3202332d558c

Additional Information

HTTP/1.1
When i choose the Page in problem and take a network diagnostic, the result is:
[aborted] Client network socket disconnected before secure TLS connection was established.

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

3

In addition, this situation just happen seriously in Auto. It isn’t serious when im using GPT-5.4 or using codex, so i do not think it as a problem of my network.

5

Hey, this is a known issue that several users have reported. There’s a main thread tracking it here: Taking longer than expected

I see you’re already on HTTP/1.1, so that’s covered. Your network diagnostic showing Client network socket disconnected before secure TLS connection was established points to connection instability between your machine and our API servers. That said, your note that it only happens on Auto, and not on GPT-5.4 or Codex directly, is interesting and useful.

A few things to try:

  • Reindex your project: File > Preferences > Cursor Settings > Indexing & Docs > Delete and Reindex.
  • Test in a brand new empty folder. Open one, start a new chat on Auto, type hello. If it works there, the issue is project-specific.
  • Are you using a VPN or proxy? If so, try without it.

The team is tracking this. Your request ID and the detail about Auto vs specific models are helpful for investigation.

Let me know how those steps go.