[Windows] Desktop app shows “ChatGPT hit a snag” and loses a long-running terminal session
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)?
26.901.51231
What subscription do you have?
Pro
What platform is your computer?
Microsoft Windows NT 10.0.26200.0 x64
What issue are you seeing?
While Codex is running a long-running terminal command, the desktop app intermittently changes to:
ChatGPT hit a snag — Something went wrong. Restart ChatGPT to try again.
https://github.com/user-attachments/assets/fd31add4-f25c-41ea-bef2-d3d39dc78b36
The active or background terminal is stopped or lost with the app session. The command's final stdout, stderr, and exit status are not returned, and there is no reliable way to determine whether the child process completed after the UI fails.
This appears to be caused by the Codex desktop session rather than user input.
What steps can reproduce the bug?
- Open a Codex desktop task with a local workspace.
- Start a terminal command expected to run longer than several seconds, such as a multi-file test or build command.
- Leave the task and terminal otherwise untouched.
- After approximately 10–15 seconds, observe the desktop window.
- The app may show the “ChatGPT hit a snag” error screen and the terminal session may disappear or be marked aborted.
- Restarting the app does not expose the interrupted command's final result.
The behavior has reproduced with both a foreground tool command and a command started as a background or interactive terminal operation.
What is the expected behavior?
The desktop app should remain available while the child process runs. If the Codex UI or terminal transport fails, the child process should either be cleanly terminated with a clear exit reason or remain independently pollable. The final stdout, stderr, exit code, and command/session state should be preserved and shown after reconnect or restart.
Additional information
The failure occurred during local software test execution; it was not caused by a user interrupt. Short commands that finish quickly return normally. A screen recording and screenshot show the generic error screen appearing during the test run; no repository files, customer data, private logs, credentials, tokens, or project-specific artifacts are attached. The practical workaround is to run long commands from an independent PowerShell or Windows Terminal session.
Suggested labels: app, bug, windows-os, tool-calls.
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 reproducing the failure in a Codex desktop task on Windows with both foreground and background terminal commands running longer than several seconds. Trace the desktop session and terminal lifecycle around the reported 10–15 second failure. Done means the app remains available or reports a clear termination, and final stdout, stderr, exit status, and session state remain available after reconnect or restart.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- cli, desktop
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100