Codex consistently escapes its sandbox
Resolved 💬 2 comments Opened Dec 3, 2025 by dagkodingentur Closed Dec 3, 2025
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of Codex is running?
codex-cli 0.64.0
What subscription do you have?
Pro
Which model were you using?
gpt-5.1-codex-max xhigh
What platform is your computer?
Windows 10 x64
What issue are you seeing?
When starting the model in a directory, the model consistently exits the sandbox by jumping one directory up.
If I start it in d:\work\project and asks it about the current working directory or asks it to run 'pwd' it reports d:\work
This bug has been for the last 4-5 versions AT LEAST. Please fix it! This creates lots of problems.
The model also gets confused and makes d:\work\project\project as it gets confused about why the project directory is missing.
What steps can reproduce the bug?
Uploaded thread: 019ae44c-345f-7720-ab63-575daa389ff4
What is the expected behavior?
_No response_
Additional information
_No response_
2 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
Thanks for the bug report. This appears to be model behavior — that is, behavior that is "baked in" to the model itself, as opposed to behavior of the codex harness or sandboxing mechanism. We'll share the feedback with the team responsible for training our codex models. We're tracking this feedback and similar reports so we can make future models more robust and efficient. And thanks for using the
/feedbackcommand. That additional information is useful.