Request ID: 90cfed45-7d02-43c6-9a7c-6970ff26b160
Connection stalled
LTe: Connection stalled
at rpf (vscode-file://vscode-app/tmp/.mount_CursorNFv3uZ/usr/share/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:9096:4764)
at fva.run (vscode-file://vscode-app/tmp/.mount_CursorNFv3uZ/usr/share/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:9096:8170)
at async Hyt.runAgentLoop (vscode-file://vscode-app/tmp/.mount_CursorNFv3uZ/usr/share/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34196:57047)
at async Zpc.streamFromAgentBackend (vscode-file://vscode-app/tmp/.mount_CursorNFv3uZ/usr/share/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34245:7695)
at async Zpc.getAgentStreamResponse (vscode-file://vscode-app/tmp/.mount_CursorNFv3uZ/usr/share/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34245:8436)
at async FTe.submitChatMaybeAbortCurrent (vscode-file://vscode-app/tmp/.mount_CursorNFv3uZ/usr/share/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:9170:14575)
Steps to Reproduce
I was just trying to prompt with 3 images of a feature’s bug I was debugging with cursor
Expected Behavior
prompt should’ve worked without giving this error persistently
Request ID: cb3fff84-55f5-4e32-b9a4-5de8db12bded
Connection stalled
LTe: Connection stalled
at rpf (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:9096:4764)
at fva.run (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:9096:8170)
at async Hyt.runAgentLoop (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34196:57047)
at async Zpc.streamFromAgentBackend (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34245:7695)
at async Zpc.getAgentStreamResponse (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34245:8436)
at async FTe.submitChatMaybeAbortCurrent (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:9170:14575)
at async Object.Ei [as onSubmit] (vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:32994:3808)
at async vscode-file://vscode-app/c:/Users/bruno/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:32968:60273
Hey, this looks like a connection issue that’s been showing up in recent versions. Here are a few things to try:
Quick fixes:
Disable HTTP/2. Add this to your settings.json (Cmd/Ctrl + Shift + P > Preferences: Open User Settings JSON):
"cursor.general.disableHttp2": true
Restart Cursor.
Check network settings: Cursor Settings > Network > Run Diagnostics. If everything is green but it still won’t work, share a screenshot of the results.
Try without images first. Send a text-only prompt in a new chat. If that works, the issue might be related to image handling.
In short: rename your project folder in your OS (for example, MyProject → MyProject_v2) and reopen it in Cursor. This resets the local cache and index for that workspace.
Downside: you’ll lose the chat history for that project, but functionality should come back.
You can also check the console for errors: Help > Toggle Developer Tools > Console. If you see anything red when you try to send a message, share a screenshot.
Let me know what worked, or if you want to dig deeper.
Cursor was a promising tool, but it’s becoming increasingly greedy. I genuinely don’t understand how AI was more affordable and useful in 2024–2025 than it is in 2026. It makes no sense.
I’m on a Pro plan — I understand that offering unlimited access to premium models isn’t sustainable. But Auto was supposed to be unlimited. That was a core part of the value proposition. At the very least, you could offer Composer 1 as a free, unlimited-token model. That would be a fair and respectable offer: if we’re paying, we should have access to at least one model without usage limits.