Codex Windows fails to launch after updates due to incomplete cua_node runtime staging
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)?
26.901.5280.0
What subscription do you have?
ChatGPT Plus
What platform is your computer?
Windows 11 Home, Version 25H2, OS Build 26200.9278, x64
What issue are you seeing?
After updating Codex, the application starts multiple ChatGPT.exe processes in the background but no visible window appears.
I reproduced this on two consecutive Codex versions:
- 26.901.5003.0
- 26.901.5280.0
The issue appears to be related to incomplete cua_node runtime staging under:
%LOCALAPPDATA%\OpenAI\Codex\runtimes\cua_node
On the first occurrence, Codex repeatedly created .staging-* directories where node.exe existed but node_repl.exe was missing.
The same launch failure returned immediately after updating to 26.901.5280.0.
What steps can reproduce the bug?
-
Install/run Codex on Windows normally.
-
Allow Codex to update to a newer version.
-
After the update completes and Codex restarts, try to launch the app.
-
Multiple
ChatGPT.exeprocesses start in the background, but no Codex window appears. -
Check:
%LOCALAPPDATA%\OpenAI\Codex\runtimes\cua_node -
Codex creates repeated
.staging-*runtime directories.
For version 26.901.5280.0, I observed:
.staging-440c4f095d41ea30-K01Wy1
.staging-440c4f095d41ea30-aIMlCB
The expected runtime hash was:
440c4f095d41ea30
What is the expected behavior?
Codex should complete the bundled cua_node runtime staging/extraction successfully after an update and then launch the application window normally.
Additional information
Repair, Reset, multiple Windows restarts, and restarting Codex did not fix the issue.
I was able to restore Codex manually by copying the complete bundled runtime from:
C:\Program Files\WindowsApps\OpenAI.Codex_26.901.5280.0_x64__2p2nqsd0c76g0\app\resources\cua_node
to:
%LOCALAPPDATA%\OpenAI\Codex\runtimes\cua_node\440c4f095d41ea30
using:
xcopy /E /I /H /Y /G
The copy completed with:
4683 File(s) copied
After the workaround:
node.exe= presentnode_repl.exe= presentmanifest.json= present
Codex then launched successfully.
The same problem previously occurred on version 26.901.5003.0 with runtime hash:
6a86821985684e13
On that occurrence I captured repeated .staging-* directories where:
node.exe = True
node_repl.exe = False
This strongly suggests a Windows runtime staging/extraction or relocation regression.
Related issue: #42412
If this reproduces again, I can capture the full pre-repair .staging-* directory with dir /s, file count, and checks for manifest.json and node_repl.exe.
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 tracing the Windows runtime staging path for %LOCALAPPDATA%\OpenAI\Codex\runtimes\cua_node and compare it with the bundled C:\Program Files\WindowsApps\...\app\resources\cua_node contents described here. Reproduce an update and inspect the .staging-* directories, checking whether node.exe, node_repl.exe, and manifest.json are all present. Done means the runtime stages completely and Codex launches a visible window without manual copying.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- rust
- Domain
- desktop, devtools, operating-systems
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100