Cursor 2.1.46 keeps freezing and is not usable

[Sepori]

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Ever since I upgraded to 2.1.46, I continue getting frequent freezing of the entire IDE. Sometimes it happens while the AI is working and then other times it happens right after the AI completes. This is happening in Ask mode, Auto mode and in Agent mode with Composer selected. Have not tested other individual agents yet with auto mode off. I did read through the forum and see this issue has been reported previously, but this is the first time I’ve experienced the issue and it has completely shut down our production coding capability. Nothing on my Windows 11 machine has changed since I updated Cursor and I don’t use VPN. Need resolution ASAP.

Steps to Reproduce

1. Open Cursor
2. Select Ask or Agent mode
3. Give Cursor a task
4. Buy a lotto ticket and see if Cursor freezes. I have not found any predictability to the freezing.

Expected Behavior

Cursor should not completely freeze up and I should not have to use Windows task manager to End Task to close Cursor.

Operating System

Windows 10/11

Current Cursor Version (Menu → About Cursor → Copy)

Cursor version 2.1.46

For AI issues: which model did you use?

Auto and Composer

Does this stop you from using Cursor

Yes - Cursor is unusable

[deanrie]

Hey, thanks for the report. It looks like a known freeze issue in 2.1.39-2.1.46 on Windows. Let’s quickly narrow down the cause:

Please attach:

• Screenshot of Settings > Network > Run Diagnostics
• Request ID from the frozen session (chat menu icon > Copy Request ID)
• Console errors: Help > Toggle Developer Tools > Console (open before reproducing)
• Screenshot of CTRL+Shift+P > Developer: Open Process Explorer (what’s eating CPU/RAM during the freeze)
• Indexing panel status (green/error?) and if there are MCP servers in Settings (temporarily disable them)

Try:

• VS Code Settings > search “HTTP/2” > enable “Disable HTTP/2”
• Disable MCP servers (if any) and restart
• Close all open tabs/files and restart (this helped here: Cursor app becomes unresponsive and crashes after opening a project)
• Check Indexing: if there’s an error/“Initializing…”, try opening a smaller subfolder of the repo (see cases: The Cursor IDE freezes at the "planning next moves" stage)

Relevant threads we’re working on:

• Final Review step/freeze: Cursor freezes on final “Review Changes” step in Agent Mode
• Planning/hanging: Cursor is unresponsive after sending a prompt to plan

Send the artifacts above - I’ll assign to engineers and prioritize. Let me know if this helps.

[Sepori]

Dean,

Your email system has bounced me back twice. First time because I’m not allowed to send more than 3 screenshots and second time because I copy/pasted your email address (or something).

Here is email part 2 of 2…

1. Screenshot of Cursor’s Developer Tools Console that was open during the first freeze. The last lines on this screenshot were the last lines that recorded

2. 

4. I didn’t have process explorer open but this is a screenshot of Task Manager about a couple seconds after second freeze happened… My dev machine has an i9 processor, 8Tb SSD with 4Gb free, and 64Gb of RAM. Cursor is 0% in this pic but it didn’t go much higher than 10% to 15% when running. I’m also running an Nvidia 4090, so hardware capacity wise, there’s no reason to freeze up.

5. 

Other notes…

1. I’m running Cursor pretty much out of the box. No MCP servers and just extensions for Github, Docker and PowerShell.

2. I disabled GPU acceleration before both freezes today

3. I was surprised at how many errors and warnings are displayed in the Developer Tools Console. This screenshot here captures just re-opening Cursor after the second freeze. There is one issue I can investigate with npm not being able to parse a file, but you guys have quite a few on your side. Health check errors and connection errors

4. 

5. Here is the text of several issues in the screenshot

6. [TerminalExecutionServiceProxy] v3 health check failed (took 6207ms)

     >  > <br>_healthCheckV3<br> | <br>@<br> | <br>workbench.desktop.main.js:9304<br> |
     > - | - | - | - |
   

7. workbench.desktop.main.js:9304 [TerminalExecutionServiceProxy] v3 health check failed, staying with current version

     >  > <br>_initializeDelegate<br> | <br>@<br> | <br>workbench.desktop.main.js:9304<br> |
     > - | - | - | - |
   

8. workbench.desktop.main.js:9304 [TerminalExecutionServiceProxy] v3 health check failed (took 1914ms)

     >  > <br>_healthCheckV3<br> | <br>@<br> | <br>workbench.desktop.main.js:9304<br> |
     > - | - | - | - |
   

9. workbench.desktop.main.js:9304 [TerminalExecutionServiceProxy] Cannot switch to v3: health check failed

     >  > <br>_updateDelegateBasedOnSettings<br> | <br>@<br> | <br>workbench.desktop.main.js:9304<br> |
     > - | - | - | - |
   

10. workbench.desktop.main.js:55
    ERR
    [Extension Host] [otel.error] {“stack”:“OTLPExporterError: Bad Request\n\tat IncomingMessage. (c:\Program Files\cursor\resources\app\node_modules\@opentelemetry\otlp-exporter-base\build\src\transport\http-transport-utils.js:52:31)\n\tat IncomingMessage.emit (node:events:531:35)\n\tat endReadableNT (node:internal/streams/readable:1698:12)\n\tat process.processTicksAndRejections (node:internal/process/task_queues:90:21)”,“message”:“Bad Request”,“code”:“400”,“name”:“OTLPExporterError”,“data”:“{"error":"Trace spans collection is not enabled for this user"}”}

      >  > <br>error<br> | <br>@<br> | <br>workbench.desktop.main.js:55<br> |
      > - | - | - | - |
    

11. workbench.desktop.main.js:6299
    [Extension Host]
    [otel.error] {“stack”:“OTLPExporterError: Bad Request\n\tat IncomingMessage. (c:\Program Files\cursor\resources\app\node_modules\@opentelemetry\otlp-exporter-base\build\src\transport\http-transport-utils.js:52:31)\n\tat IncomingMessage.emit (node:events:531:35)\n\tat endReadableNT (node:internal/streams/readable:1698:12)\n\tat process.processTicksAndRejections (node:internal/process/task_queues:90:21)”,“message”:“Bad Request”,“code”:“400”,“name”:“OTLPExporterError”,“data”:“{"error":"Trace spans collection is not enabled for this user"}”}

12. workbench.desktop.main.js:6995 [transport] Connect error in unary AI connect ConnectError: [unauthenticated] Error at t (workbench.desktop.main.js:6331:410306) at async Object.getTeamCommands (workbench.desktop.main.js:565:74387) at async rTo.loadTeamCommands (workbench.desktop.main.js:8980:2992)

13. Object1. workbench.desktop.main.js:8980 Failed to load team commands: ConnectError: [unauthenticated] Error at t (workbench.desktop.main.js:6331:410306) at async Object.getTeamCommands (workbench.desktop.main.js:565:74387) at async rTo.loadTeamCommands (workbench.desktop.main.js:8980:2992)

14. workbench.desktop.main.js:9304 [TerminalExecutionServiceProxy] v3 health check failed (took 547ms)

      >  > <br>_healthCheckV3<br> | <br>@<br> | <br>workbench.desktop.main.js:9304<br> |
      > - | - | - | - |
    

15. workbench.desktop.main.js:9304 [TerminalExecutionServiceProxy] Cannot switch to v3: health check failed

      >  > <br>_updateDelegateBasedOnSettings<br> | <br>@<br> | <br>workbench.desktop.main.js:9304<br> |
      > - | - | - | - |
    

Please let me know if any of this helps, and I will report back on stability after trying to resolve the npm error about not parsing the package.json file.

-SC

[Sepori]

Dean,

Here is email part 1 of 2…

Thank you so much for the response today. Below are as many things as I could capture from your list…

1. Screenshot of network Diagnostic. I ran this right after a freeze happened, but I had not yet changed the HTTP/2 setting. That said, it passed.

2. 

3. Here is the diagnostic with HTTP/1.1 setting. Ran this after the second freeze when I was on the 1.1 setting.
   

   

4. The freeze happened twice so far this evening. The first time I was in Ask mode and the request had just completed when it froze. Had to use Task Manager to End Task. Re-opened cursor and my original prompt was still visible, but the results I just read were gone. Request ID when freeze happened…13200a20-68d7-4371-95ad-563871d07ac3

5. The second freeze happened in Agent mode and the request had just completed. Here is the request ID and screenshots below… 604dd1f7-81d1-4c43-92b5-6a8f3bb9f606

6. Here is a screenshot when the screen was frozen…
   

   

See part 2 of 2 in next email…

-SC

[deanrie]

Thanks, this is very helpful. I see:

• OTLPExporterError 400 - not critical, can be ignored
• Repeated “TerminalExecutionServiceProxy v3 health check failed”
• npm errors (“failed to parse package.json”, “Timed out getting tasks from npm”)
• Massive read_file/search calls during the freeze

What else we need:

• Screenshot of CTRL+Shift+P > Developer: Open Process Explorer during the freeze
• Screenshot of the Indexing tab (error?). Try opening a small subfolder/empty project - does the freeze persist?
• Are you using network share/SSH/WSL? If yes - check the same project locally on disk

Please try these short steps:

• Close all open tabs/files and restart
• Run without extensions: cursor --disable-extensions and test in a clean window

We’re tracking a similar bug: Cursor freezes on final “Review Changes” step in Agent Mode. I’ll attach your Request ID and logs to the ticket. Let me know if the behavior changes after closing tabs/testing on a small folder.

[Sepori]

Dean,

I updated to 2.1.47 and ran it for almost 12 hours yesterday with zero freezes. I’m not sure what you guys did, but please let the engineering team know not to change it!

Accordingly, I have not been able to capture any additional freeze details for you. There are a few quirks I found with trying to open plans after created, but also have a work around.

I just saw another version notification this am and I’m hesitant to make any changes. Is there a way to see release notes for minor versions? I know you guys have the major versions on your website. Is it safe to update to 2.1.48, and if so what changed?

-SC