macOS][26.908.40834] Codex Desktop consistently hangs on the second prompt while CLI works normally
Open
Nobody has claimed this yet.
app
bug
session
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.4k
- PR merge metrics
- PR metrics pending
Description
What version of Codex CLI is running?
codex-cli 0.154.0-alpha.6.2
What subscription do you have?
Pro
Which model were you using?
No response
What platform is your computer?
MacOS 27
What terminal emulator and version are you using (if applicable)?
No response
Codex doctor report
### What version of the Codex App are you using?
Codex Desktop: 26.908.40834
Codex CLI: 0.154.0-alpha.6.2
### Subscription
ChatGPT Pro
### Platform
Apple Silicon iMac M1
macOS 27.0
### Issue
Codex Desktop consistently works for the first prompt in a new conversation, but the second prompt hangs indefinitely.
This is reproducible in new conversations.
The first prompt completes normally.
The second prompt remains stuck and never produces a response.
Codex CLI works correctly on the same machine, using the same ChatGPT account.
### Steps to reproduce
1. Launch Codex Desktop.
2. Start a new conversation.
3. Send:
"Responde apenas: PRIMEIRO OK"
4. The first response completes successfully.
5. Send:
"Responde apenas: SEGUNDO OK"
6. The second turn hangs indefinitely.
This behavior reproduces repeatedly.
### Expected behavior
The second and subsequent prompts should execute normally.
### Actual behavior
The first turn succeeds, but the second turn hangs indefinitely.
### CLI verification
The bundled CLI works correctly:
codex-cli 0.154.0-alpha.6.2
Authentication:
Logged in using ChatGPT
Direct CLI execution succeeds:
/Applications/Codex.app/Contents/Resources/codex exec --skip-git-repo-check "Responde apenas: CODEX OK"
Result:
model: gpt-5.6-sol
provider: openai
approval: never
sandbox: danger-full-access
reasoning effort: medium
CODEX OK
Therefore the account, authentication, backend connection, and Codex execution engine appear to be working normally.
### Troubleshooting already performed
- Upgraded macOS to macOS 27.
- Fully quit and restarted Codex.
- Removed forced model configuration from ~/.codex/config.toml.
- Removed:
model = "gpt-6-astra"
model_reasoning_effort = "high"
service_tier = "priority"
- Recreated the Codex Desktop application state by renaming:
~/Library/Application Support/Codex
- Reset macOS preferences:
~/Library/Preferences/com.openai.codex.plist
- Restarted cfprefsd.
- Confirmed no queue_1.sqlite / queue*.sqlite files exist.
- Verified authentication through Codex CLI.
- Verified direct CLI prompts work correctly.
None of these steps fixed the Desktop behavior.
### Logs
During the hang there is no obvious application crash or network failure.
The macOS unified log repeatedly shows preference lookups such as:
found no value for key <private> in CFPrefsSearchListSource
Domain: com.openai.codex
These messages repeat while the second prompt remains stuck.
There are also some securityd / Keychain messages, including:
Error Domain=NSOSStatusErrorDomain Code=-50
"no object for key query"
However, Codex CLI continues to work normally.
### Additional observation
The issue appears isolated to the Codex Desktop session/thread continuation layer rather than to the model backend, authentication, or network connection.
A fresh Desktop state still reproduces the issue:
first turn succeeds -> second turn hangs.
### Reproducibility
100% on my machine.
What issue are you seeing?
The issue is reproducible in new conversations:
- First prompt completes normally.
- Second prompt gets stuck and never returns a response.
- Codex CLI works correctly on the same Mac and same ChatGPT account.
- Authentication is valid.
- Recreating the Codex Desktop application state and resetting com.openai.codex preferences did not fix it.
So the issue appears isolated to Codex Desktop conversation/session continuation, not the backend, account, or CLI.
What steps can reproduce the bug?
- Open Codex Desktop on macOS.
- Start a new conversation.
- Send a first prompt, for example:
Reply only: FIRST OK - Wait for the first response to complete successfully.
- Send a second prompt in the same conversation:
Reply only: SECOND OK - The second prompt hangs indefinitely and never returns a response.
- Repeat in another new conversation: the same pattern occurs again.
For comparison, running the bundled Codex CLI on the same Mac and same ChatGPT account works normally.
What is the expected behavior?
No response
Additional information
No response
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 second-turn hang in Codex Desktop, then compare it with the bundled /Applications/Codex.app/Contents/Resources/codex exec entry point, which works normally. Check whether resetting ~/Library/Application Support/Codex and com.openai.codex.plist changes the session behavior. Done means the second and later prompts complete reliably in new Desktop conversations on macOS.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- macos, rust
- Domain
- desktop
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100