[Bug] Model generating incorrect bash commands and false success reports

Bug Description
Claude was doing well up until about an hour ago where it started utterly messing up bash commands, not writing things properly, it was like something happend. I have never experience this level of terrible with claude code and the models. it was kicking off parallel bash things that caused locking issues and errors which it could write then documents. it was telling me it did things and were successful but it wasn't. very bad experience tonight

Environment Info

• Platform: darwin
• Terminal: ghostty
• Version: 2.1.158
• Feedback ID: 32830218-9c1b-4809-bbb0-fda16dc251f0

Errors

[{"error":"Error: Failed to submit feedback:403\n    at fo8 (/$bunfs/root/src/entrypoints/cli.js:6614:2263)\n    at async Mk6 (/$bunfs/root/src/entrypoints/cli.js:6614:3538)\n    at async <anonymous> (/$bunfs/root/src/entrypoints/cli.js:6614:7432)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-05-31T01:18:30.329Z"},{"error":"Error: Failed to submit feedback:403\n    at fo8 (/$bunfs/root/src/entrypoints/cli.js:6614:2263)\n    at async Mk6 (/$bunfs/root/src/entrypoints/cli.js:6614:3538)\n    at async <anonymous> (/$bunfs/root/src/entrypoints/cli.js:6614:7432)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-05-31T01:19:14.772Z"}]