Where does the bug appear (feature/product)?
Grok Bot
Describe the Bug
I pay one hundred dollars a month for SuperGrok Plus. For the past twenty-four hours my Grok Bot has been unable to run any commands whatsoever. The box is locked. I have hourly routines that have run every single day that are now completely broken, and I can’t open or manage them. This isn’t a minor glitch — the product I subscribed to is non-functional.
Here’s what happened. I noticed the box was at around seventy-five gigabytes of consumed disk space and proactively asked my Grok Bot agent to identify unused files. It reported seventy-three gigabytes that were no longer needed. I authorized the deletion. It confirmed all seventy-three gigabytes were deleted. After that, the agent stopped responding to commands — responses took forever. I asked it to diagnose itself, and it said the box seemed to need a reset and sent me a link labeled “update your computer.” On mobile iOS, that link opened a settings page offering only a “reset your computer” option. I was hesitant because I feared losing data, but the bot assured me it wouldn’t delete anything — it was just an update — and sent me the same link again. In reality, that link triggered a rebuild, not an update. After hours of the bot not responding to any commands, I went ahead and reset the computer to try to release the disk space freed by the seventy-three gigabyte deletion. After the reset, it appears the system restored from a previous snapshot taken when disk space was critically low, and rebuilt the VM from that state. Since then, not a single command can run on the box or the VM.
This seems like a critical architectural flaw in Grok Bot’s implementation. The system should never allow a VM to reach critically low disk space to the point where it can’t execute any commands at all. There should be proactive disk-space monitoring, automated cleanup thresholds, and snapshot management that doesn’t restore a broken state after a reset.
Steps to Reproduce
Reach approximately seventy-five gigabytes of disk consumption on a Grok Bot VM. Have the agent identify and delete seventy-three gigabytes of unused files. Attempt to run any command — the agent becomes unresponsive. Use the “update your computer” link, which actually triggers a rebuild. After reset, the VM restores from a critically low disk-space snapshot. No commands execute.
Expected Behavior
Grok Bot should monitor disk usage and prevent the VM from hitting critical thresholds. Deletions should be reflected in the live state, not a stale snapshot. A reset or rebuild should restore from a healthy snapshot, not one captured during a disk crisis. Command execution should never be fully unavailable.
Screenshots / Screen Recordings
Operating System
Linux
iOS
Other
Version Information
Grok Bot Latest iOS app update as of Oct. 28th
For AI issues: which model did you use?
Not applicable — this is an infrastructure lockout, not a model output problem
For AI issues: add Request ID with privacy disabled
f9a7046a-279b-47e5-ab48-6e8dc12daba1
Additional Information
SuperGrok Plus subscriber, one hundred dollars monthly. Multiple agents affected. Support emails to both Cursor and xAI have gone unanswered. The seventy-three gigabytes of deleted files are gone from the live disk but the restored snapshot predates the cleanup, so that space is effectively unrecoverable through the current reset path. This has been ongoing for a full day with zero resolution path from my end.
Does this stop you from using Cursor
Yes - Cursor is unusable