codex doctor falsely reports healthy state database corruption inside managed workspace sandbox #33145

What version of Codex CLI is running?

Codex CLI 0.144.4

What subscription do you have?

plus

Which model were you using?

sol-ultra

What platform is your computer?

debian 13

What terminal emulator and version are you using (if applicable)?

kitty, herdr

Codex doctor report

019f6149-7a69-7351-8c02-101b4288b429

What issue are you seeing?

Codex CLI 0.144.4 on Debian 13, using remote control and the built-in
:workspace permission profile.

What steps can reproduce the bug?

Uploaded thread: 019f6149-7a69-7351-8c02-101b4288b429

What is the expected behavior?

Steps

1. Start Codex with the built-in :workspace permission profile.
2. From an agent tool call, run:
   codex doctor --summary --no-color
3. Observe the state-database corruption and stale app-server warnings.
4. Run codex doctor --json through an approved host-level command.
5. Observe that all SQLite integrity checks, app-server checks, HTTP, and
   WebSocket checks pass without any repair between runs.

Expected

Doctor should classify checks it cannot perform because of sandbox restrictions
as unavailable or sandbox-constrained. It must not diagnose corruption or
recommend moving a healthy database aside.