Enabling unified_exec in config.toml causes all shell commands to fail (exit code -1073741502)

What version of Codex is running?

codex-cli 0.63.0

What subscription do you have?

Team

Which model were you using?

gpt-5.1

What platform is your computer?

Microsoft Windows NT 10.0.22631.0 x64

What issue are you seeing?

After I enable unified_exec in my Codex config, all shell commands triggered from Codex stop working.

I added this to %USERPROFILE%\.codex\config.toml:

[features]
unified_exec = true

After that, any command Codex tries to run via the integrated shell fails immediately with exit code -1073741502 (Windows error 0xC0000142 – process initialization failed), with no stdout/stderr output at all.

This happens regardless of the actual command:

pwsh -NoLogo -Command "node -v"

Before enabling unified_exec, the same commands executed via Codex worked normally.

From a normal PowerShell window on the same machine, these commands work fine, so it’s not an OS or Node/curl issue; it only breaks when Codex is using unified_exec.

So effectively, enabling unified_exec = true makes Codex’s command execution backend unusable on my Windows machine.

If needed, I can provide the relevant excerpts from codex-tui.log, but currently the UI just shows something like “Process exited with code -1073741502” and empty output.

What steps can reproduce the bug?

On Windows, create or edit %USERPROFILE%\.codex\config.toml and add:

[features]
unified_exec = true

In a project, ask the assistant to run a simple command, e.g.:

pwsh -NoLogo -Command "Get-Location"
or pwsh -NoLogo -Command "node -v"

The command exits with code -1073741502.
No stdout/stderr is shown, even for very simple commands

What is the expected behavior?

_No response_

Additional information

When an error occurs, codex-tui.log does not have new logs