Codex Desktop intermittently resets active conversation to splash screen and eventually crashes with SIGTRAP on macOS
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.4k
- PR merge metrics
- PR metrics pending
Description
What version of the Codex App are you using (From “About Codex” dialog)?
26.908.40834 (8881)
What subscription do you have?
ChatGPT Pro
What platform is your computer?
Darwin 25.6.0 arm64 arm
What issue are you seeing?
Codex Desktop intermittently resets the active conversation and shows the OpenAI/Codex splash screen. When this happens, I am taken out of the currently active conversation.
This can happen even when no coding task is running. I can reproduce it by simply opening and switching between conversations.
Opening a new window temporarily works around the problem, but after continued use the application can eventually crash completely.
During one UI reset, the macOS unified log showed:
RemoteHostedPIPContent active thread changed threadID=<private>
RemoteHostedPIPContent active thread changed threadID=(null)
RemoteHostedPIPContent active thread changed threadID=<private>
The main ChatGPT process and codex app-server remained alive during the UI reset.
I also captured a full application crash after opening a new window. The macOS crash report shows:
Exception Type: EXC_BREAKPOINT (SIGTRAP)
Triggered by Thread: 0 CrBrowserMain
Termination Reason: SIGNAL 5 Trace/BPT trap: 5
The crash occurred inside Codex Framework / Chromium on the CrBrowserMain thread.
This does not appear to require Browser, Computer Use, or an active coding task to reproduce.
What steps can reproduce the bug?
- Launch Codex Desktop on macOS.
- Open an existing conversation.
- Switch between several conversations normally.
- Continue using the app or keep switching conversations.
- Intermittently, the current window displays the OpenAI/Codex splash screen and exits the active conversation.
- Open a new window with Cmd+N.
- The new window initially works normally.
- After continued use, the application may eventually crash completely.
No coding task needs to be running to reproduce the conversation reset.
The issue also reproduces without using Browser or Computer Use.
I have not identified a deterministic number of conversation switches or amount of time required before the issue occurs.
No specific session appears to trigger the issue. It occurs across conversations and can reproduce while only switching between them.
What is the expected behavior?
Switching between conversations should keep the selected conversation active without resetting the window to the splash screen.
The Codex window should remain usable without requiring a new window to recover, and normal conversation navigation should not cause the application to crash.
Additional information
I performed the following troubleshooting without resolving the issue:
- Cleared Chromium/renderer caches, including Cache, Code Cache, GPU-related caches, GraphiteDawnCache, and GPUPersistentCache.
- Reset
browser-sidebar-page-states.json; the file was regenerated, but the issue continued. - Reset
Default/Session Storage; the issue continued. - Tested without Browser / Computer Use; the issue still reproduced.
- Reproduced the issue while only switching conversations, without running a coding task.
- Verified that the ChatGPT main process and
codex app-serverremain alive during the splash-screen/conversation-reset behavior. - Opening a new window temporarily works around the UI reset.
Around observed conversation resets, macOS logs also contained AppKit state restoration activity such as:
NSPersistentUIManager flushAllChanges
NSPersistentUIManager flushAllChanges finishing enqueued operations
NSPersistentUIManager flushAllChanges_block_invoke writing records
I also captured a macOS .ips crash report from a subsequent full application crash.
Crash summary:
Codex App: 26.908.40834 (8881)
Architecture: ARM64
Exception Type: EXC_BREAKPOINT (SIGTRAP)
Triggered Thread: CrBrowserMain
Termination: Trace/BPT trap: 5
Codex Framework: 152.0.7977.83
The complete macOS .ips crash report is attached.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing the attached macOS .ips crash report and reproducing the listed conversation-switching steps on macOS. Use the unified-log entries and crash summary to compare the splash-screen reset with the later SIGTRAP crash. Done means conversation switching no longer resets the window or causes the application to crash.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- macos
- Domain
- desktop
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100