I have had no access to my Ultra subscription since 7am this morning - just getting ‘Tool call error’
I have logged a call with support, and initially Sam tried to help with standard troubleshooting, but for the last 5 hours, I’ve had no response when asking for updates etc. Compete radio silence. I’d expect more for a $200 per month subscription! I have asked for SLA’s but that request has been ignored too. Does anyone have any experience with getting support from Cursor? Do they have an established support function, or is it a “best effort” approach? Really considering cancelling, as I have missed a critical deadline as a result of the service outage. Sonnet 4.5, Grok, ChatGPT etc. ALL MODELS affected - Complete service outage
Having to log for a second time, as my original ticket was deleted for some reason
Steps to Reproduce
Any command with any model fails. Connected to different networks in different buildings, so NOT a local network issue.
Expected Behavior
I expect access to the IDE!
Operating System
MacOS
Current Cursor Version (Menu → About Cursor → Copy)
Sonnet 4.5, Grok, ChatGPT etc. ALL MODELS affected - Complete service outage
For AI issues: add Request ID with privacy disabled
Request ID: a2a5e8bf-56d4-49a5-bfed-0298d1fa0f51
{“error”:“ERROR_BAD_REQUEST”,“details”:{“title”:“Tool call error”,“detail”:“Failed to read file.”,“additionalInfo”:{},“buttons”:,“planChoices”:},“isExpected”:true}
ConnectError: [invalid_argument] Error
at QJc.$endAiConnectTransportReportError (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:4989:399336)
at Ywo._doInvokeHandler (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:36007)
at Ywo._invokeHandler (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:35749)
at Ywo._receiveRequest (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:34514)
at Ywo._receiveOneMessage (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:33336)
at GDt.value (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:31429)
at _e._deliver (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:49:2962)
at _e.fire (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:49:3283)
at ggt.fire (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:4974:12156)
at MessagePort. (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:7498:18439)
Also, I logged a call with support at 7am (10+ hours ago) and besides the basic troubleshooting that Sam did in the first couple of hours, I have heard nothing back from anyone, despite multiple emails requesting an update on the situation. I also requested SLA’s for Ultra subscription, but this request has also been ignored - how do I escalate a complete service outage? I am literally not getting what I am paying $200 per month for!
Hey, thanks for the report. This is a known issue that the engineering team is actively working on. The error you’re seeing (ERROR_BAD_REQUEST: "Failed to read file") matches reports from other users.
Please try these workarounds:
Start a completely fresh chat (not just restart Cursor), and avoid reopening the failed chat, since this can cause code rollbacks.
Check if MCP servers are enabled: go to Cursor Settings > Features > MCP and temporarily disable any MCP servers if present.
Try opening a different workspace or project to see if the issue is specific to your current workspace.
Check if there’s a .cursorignore file in the root of your workspace.
Important: use version control (git) to protect your work while you’re experiencing this issue.
Please let me know if any of these steps help, and whether you have MCP servers enabled or a .cursorignore file in your workspace.
Regarding support SLAs, I’ll share your feedback with the team about response times for Ultra subscribers.
Thanks for getting back to me. In answer to your questions:
Start a completely fresh chat (not just restart Cursor), and avoid reopening the failed chat, since this can cause code rollbacks.- this is not possible, as I am working on an iterative code development stage, so if I start a completely fresh chat it will lose the context. This will cost me about 1.5 weeks of work, which I can’t afford to lose due to a deadline.
Check if MCP servers are enabled: go to Cursor Settings > Features > MCP and temporarily disable any MCP servers if present. - Playwright was enabled - I disabled it, restarted Cursor, but the same error persisted.
Try opening a different workspace or project to see if the issue is specific to your current workspace. - it is just the current workspace - I ran a command successfully in a workspace from the previous iteration.
Check if there’s a .cursorignore file in the root of your workspace. - no cursorignore file
I guess another way to enable me to continue with my work, is to provide a way for me to transfer the context from my existing chat to a new chat. Is this possible?
MESSAGE for the Cursor Community - Anyone else facing the same “Tool call error” that has stopped my development work for the last 36-hours whilst the support team have “apparently” been investigating this “known issue” - I have found a way to circumvent the issue:
go back in the conversation to the last successful interaction, and issue the command again, selecting ‘Revert and continue’ - this will clear the error in the subsequent command, and resume from the previous command
This is certainly better than losing the entire context window and all the work that you have done - this was the only suggested fix from Cursor support!
