Incorrect full-auto description in `codex --help` and `codex exec --help`

What version of Codex is running?

0.57.0

What subscription do you have?

ChatGPT

Which model were you using?

_No response_

What platform is your computer?

Darwin 24.3.0 arm64 arm

What issue are you seeing?

The current description of full-auto when running codex --help and codex exec --help is as follows:

      --full-auto
          Convenience alias for low-friction sandboxed automatic execution (-a on-failure, --sandbox workspace-write)

However when looking at the code inside codex-rs/tui/src/lib.rs it seems that full-auto is on-request rather than on failure:

pub async fn run_main(
    mut cli: Cli,
    codex_linux_sandbox_exe: Option<PathBuf>,
) -> std::io::Result<AppExitInfo> {
    let (sandbox_mode, approval_policy) = if cli.full_auto {
        (
            Some(SandboxMode::WorkspaceWrite),
            Some(AskForApproval::OnRequest),
        )

which is also aligned with what is written in https://github.com/openai/codex/blob/main/docs/sandbox.md#platform-sandboxing-details

What steps can reproduce the bug?

Run codex --help and codex exec --help

What is the expected behavior?

_No response_

Additional information

_No response_