There is no response in the chat / reconnecting

[Sleepzzzy]

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When I restart and make an instant request, the AI responds. However, when I make a second request, I get an endless “Reconnecting” message, which sometimes changes to “Planning next moves” and then nothing happens.

Steps to Reproduce

It’s hard to say how it’s being reproduced, because it started today. Log in to the IDE, make a request to AI, agent, auto.

After running the HTTP/2 diagnostics, there are either errors on the API, Ping, Chat, and Agent
Error: ConnectError: [unavailable] getaddrinfo EAI_AGAIN agentn.global.api5.cursor.sh

or sometimes an error in Ping, Chat, and Agent
Error: ConnectError: [aborted] read ECONNRESET

There are no errors in HTTP/1.1, but there are no changes in the flow bug.

Expected Behavior

AI will respond to requests

Screenshots / Screen Recordings

Operating System

Windows 10/11

Version Information

Version: 2.4.28 (system setup)
VSCode Version: 1.105.1
Commit: f3f5cec40024283013878b50c4f9be4002e0b580
Date: 2026-02-03T00:56:18.293Z
Build Type: Stable
Release Track: Default
Electron: 39.2.7
Chromium: 142.0.7444.235
Node.js: 22.21.1
V8: 14.2.231.21-electron.0
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use?

It appeared on Auto, but there is no difference.

Additional Information

I’ve tried reinstalling, changing versions, and switching from HTTP/2 to HTTP/1.1, turning off the VPN, but nothing has worked.

I received answers 3-4 times a day, but then the bug returned. It responds to simple requests with HTTP/1.1, but freezes on more complex requests.

Does this stop you from using Cursor

Yes - Cursor is unusable

[Sleepzzzy]

[deanrie]

Hey, thanks for the report. I can see in the console that the issue is broader. There are timeouts to api3.cursor.sh too, and ECONNRESET on all transport connections. So it’s not just DNS, the connections are being actively dropped.

A couple questions:

• Who’s your ISP?
• Do you have any antivirus or firewall installed (Kaspersky, ESET, Windows Defender with extra rules)?
• Are you on a corporate network or at home?

Please try:

1. Temporarily disable your antivirus/firewall. They often scan HTTPS traffic and can drop connections.

2. Test using a mobile hotspot. If it works there, it’s definitely a network or ISP issue.

3. Add firewall exclusions for:

   • *.cursor.sh
   • *.cursor.com

Based on the errors, something between you and Cursor’s servers is actively dropping TCP connections. It’s usually security software or your ISP.

Let me know the hotspot result, it’s the fastest way to narrow down where the issue is.

[Sleepzzzy]

Hey!
Answers to questions:

• MTS (RU ISP);
• Nope;
• I’m at home, but I’m connected to the corporate network, and when I disconnect it, the result doesn’t change.

Disabling the firewall didn’t help, and I don’t think an exception will help either.
The network settings haven’t changed either yesterday or today, but it only stopped working today.

[Sleepzzzy]

Here are more logs if it helps

[deanrie]

The logs confirm that all TCP connections to api3.cursor.sh are being forcibly dropped with ERR_CONNECTION_RESET. This is happening at the network level, not inside Cursor.

This pattern looks like Deep Packet Inspection (DPI), which MTS uses. The problem is that something between your computer and our servers is actively cutting these connections.

The key test is a mobile hotspot. This is the only way to 100% confirm the issue is on the ISP side. Please try:

1. Turn on internet sharing from your phone (mobile data, not WiFi from your router)
2. Connect your computer to the hotspot
3. Open Cursor and make a couple of requests

If it works over the hotspot, that confirms MTS is blocking the traffic. In that case, the only solution is a VPN that works (some VPNs get blocked too).

[Sleepzzzy]

I still don’t understand what the final reason was, but I have two PCs at home that are connected to the same IPS. On the second PC, Cursor worked without any issues. I swapped the wires in the router, and the requests started working on both devices. However, I had already restarted the router three times without success.

If anyone encounters this issue, try restarting the router or changing the location of the PC’s connection to the router. Although it may seem strange, it could potentially resolve the issue.

And also try to restart the PC and the router. And before turning on the VPN, check the cursor functionality, if everything is fine, then already after that connect the VPN or corporate network.