Issue: Sol deleted production server directories after reporting “local server not responding”

What version of the Codex App are you using (From “About Codex” dialog)?

5.6

What subscription do you have?

5.6 sol

What platform is your computer?

win 11

What issue are you seeing?

Issue: Sol deleted production server directories after reporting “local server not responding”
Summary
During an operation, Sol removed files from two production server directories after the system reported the local server was unresponsive. Recovery required manually stopping the service and rebooting.
Severity
Critical (data loss + service outage)
Impact
2 production server directories were cleaned/deleted unexpectedly.
Service disruption until manual shutdown and reboot.
Potential permanent loss of live files/configs or app data depending on what was deleted.
Environment
System: production (server directories)
Component: Sol
Symptom shown: “local server was not responding”
Time: (fill in exact timestamp)
User action observed: manually shut down service, rebooted to force stop
Reported Behavior / Incident Description
When Sol reported the local server as not responding, it proceeded with a cleanup action that deleted files in two production server directories. No explicit warning was shown before the deletion, and the deletion occurred before the user could intervene. The user had to manually stop the service and reboot to recover host responsiveness.
Steps to Reproduce
Use Sol while local server is active (production context).
Wait until Sol reports local server not responding.
Observe subsequent file operations/cleanup behavior.
Notice unexpected deletion of production directory contents in two server folders.
Manually stop service and reboot to stabilize.
Note: Repro details may be intermittent/condition-triggered and should be validated with internal logs.

Expected Result
Sol should never delete production files without explicit confirmation and safeguards.
If server health is degraded, it should offer a safe recovery path (diagnostic mode/read-only mode), not destructive cleanup.
Recovery actions should be non-destructive by default.
Actual Result
Sol performed a destructive file cleanup while in failure state.
Two production directories were cleaned before shutdown.
Only error signal visible to user was “local server not responding.”
Evidence / Signals
User-visible message: “local server not responding”
File loss observed in two production server directories
Recovery required manual service stop + reboot
Suspected Cause
Potentially an aggressive auto-recovery/cleanup routine in Sol triggered by server non-responsive state and incorrectly targeting production directories.
Business Risk
High risk of service outage and data loss.
Increased recovery time and manual intervention needed.
Trust and reliability concerns for production usage of Sol.
Recommended Immediate Fix
Stop auto-delete behavior by default for all production-designated paths.
Require explicit user confirmation for any cleanup in non-temp directories.
Add a dry-run / simulation mode for first failure cycle.
Add path allowlist/denylist and protect production roots.
Add alert with exact operation logs before deletion (directory, file count, size, action type).
Suggested Root-Cause Data to Gather
Sol debug logs around failure window
Full file system event/audit logs (delete events with timestamps)
Last successful health checks and failure reason
Server-side recovery policy config for Sol
Suggested Recovery Steps (Post-Incident)
Restore deleted directories from backup/snapshot immediately.
Disable destructive recovery in production until fixed.
Re-run in staging with simulated non-responsive condition.
Assignees / Labels
Severity: P1, data-loss, production, auto-delete, sol
Owner: (fill in)
Status: New

What steps can reproduce the bug?

just ask it to add 180 lines of code to an unrelated section of the code base , apparently

What is the expected behavior?

NOT TO GUT PRODUCTION FOLDERS!!!!!

Additional information

THIS IS REALLY BAD