Cursor freezes/crashes when attempting to use an MCP server

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When cursor AI agent attempts to use MCP servers after the marketplace update it will freeze the whole application.

Steps to Reproduce

Enable MCP servers
Tell AI agent to use them
Cursor freezes

Expected Behavior

To be able to use MCP servers without cursor freezing

Screenshots / Screen Recordings

Operating System

Windows 10/11

Version Information

Version: 2.5.17 (user setup)
VSCode Version: 1.105.1
Commit: 7b98dcb824ea96c9c62362a5e80dbf0d1aae4770
Date: 2026-02-17T05:58:33.110Z
Build Type: Stable
Release Track: Default
Electron: 39.3.0
Chromium: 142.0.7444.265
Node.js: 22.21.1
V8: 14.2.231.22-electron.0
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use?

Auto but issue happens on any model.

Additional Information

This particular test was done on win 11 but it also occurs on my linux laptop also. It also does not matter if I am using MCP servers added by a plugin or if I add them myself this update just straight up broke MCP for me. I have also attempted a full clean reinstall by uninstalling and going to:
My users folder and deleting .cursor folder
Going to %appdata% and deleting Cursor folder
Then downloading a fresh copy of cursor and no change, the issue still persists.

I have node v24.13.0 installed incase that is relevant.

Does this stop you from using Cursor

Yes - Cursor is unusable

Hey, thanks for the report.

A couple things that stand out:

  1. I see several MCP servers. Try turning them all off, then turn them on one by one to find which one causes the freeze, or whether it only happens with a specific combination.

  2. To pinpoint the cause, we’ll need developer logs. Right before you reproduce the freeze:

    • Open Help > Toggle Developer Tools > Console tab
    • Reproduce the issue, then take a screenshot of the console output, or copy the red errors
  3. If you can get a Request ID before it freezes: open the chat menu (three dots in the top right) > Copy Request ID.

The team is aware of a related issue where MCP tool calls can cause freezes in certain scenarios.

As a temporary workaround, try reducing the number of active MCP servers to only the ones you actually need for the current task.

Let me know what the logs show.

Looks like it is the astro docs MCP server. That is weird I have been using it for weeks without issue and now suddenly it is the issue.

Thanks for the logs and the gist.

The OTLPExporter errors in the console (“Trace spans collection is not enabled for this user”) are most likely just background noise, not the main cause of the freeze. This error is related to telemetry and by itself shouldn’t hang the app on 2.5.17.

A few things to look at:

  1. Node.js v24: This is a very new version. MCP servers run via Node, and incompatibility with the new runtime can cause issues. Try downgrading to Node v22 LTS and see if the freeze still happens.

  2. Astro-docs: This is a third-party MCP server. It might be returning a very large response that triggers the freeze. Try opening a terminal and running it manually to check how much data it returns.

  3. If the freeze happens consistently, try to grab the Request ID before everything hangs (three dots in the top-right of the chat > Copy Request ID). This will help us investigate on the server side.

Let me know how it goes.

  1. Didn’t make a difference for me
  2. I do not know how to do this sorry. Is there a guide anywhere for how to do this?
  3. It happens way too fast for me to click

I’ve confirmed it on my side, I can reproduce it. I’ve passed it on to the team.

I’ll update the thread when there’s news.

This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.