macOS Input Bug: Control-F cursor movement is broken in the Codex app
Resolved 💬 28 comments Opened Mar 7, 2026 by SoraTakai Closed Apr 10, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor)
responded on this thread — see the highlighted reply below.
What version of the Codex App are you using (From “About Codex” dialog)?
26.305.950 (863)
What subscription do you have?
Plus
What platform is your computer?
Darwin 23.6.0 arm64 arm
What issue are you seeing?
This report concerns cursor behavior in the text input field.
As you all know, under macOS, text field navigation follows Emacs-style key bindings by default:
- Control-F = move forward one character
- Control-B = move backward one character
- Control-P = move up one line
- Control-N = move down one line
- Control-A = go to the beginning of line
- Control-E = go to the end of line
Problem:
- ❌ Control-F = move forward one character ← BRINGS UP AND SHIFTS FOCUS TO "FIND" SEARCH BAR
- ✅ Control-B = move backward one character
- ✅ Control-P = move up one line
- ✅ Control-N = move down one line
- ✅ Control-A = go to the beginning of line
- ✅ Control-E = go to the end of line
This is standard default navigation behavior on macOS and should be respected.
Please fix Control-F navigation in the text input field.
Plus, you already have "Find" bound to Command-F (standard behavior and sensible); and it does NOT make sense to redundantly bind it to Control-F, which is something reserved for navigation.
What steps can reproduce the bug?
- Open Codex App
- Press Command-V (paste) something containing multiple lines of text
- Press Control-A to go to the beginning of the line
- Press Control-F - nothing happens (cursor stays there as opposed to expected forward-char cursor movement)
- (Press Control-F)+ ad infinitum - nothing happens, ever
What is the expected behavior?
Control-F at text input field → move the cursor by 1 char (if there is some text after the cursor on the same line)
Additional information
_No response_
28 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
No, it is not a dup - not even close. Thanks for your attempt nevertheless.
Anyone taking a look? Why does it take so long for anyone to even respond?
If there's no one, just use one of your models - Claude Code does this automatically.
@SoraTakai can you also include ctrl+w and ctrl+u in this issue since these commands help delete word or whole sentence?
@stym06 Both ctrl+w and ctrl+u do nothing under my environment (macOS).
I do not define any non-standard key handling via Hammerspoon or Karabiner or whatever, on Codex App.
So everyone should be getting this behavior on macOS, unless they change its default Emacs-like navigation keys.
In my current version (
26.325.21211 (1255)), pressing Control-F now opens the search bar. However, this action is already available via Command-F.Ideally, it would be great if macOS standard keybindings could be respected. Alternatively, having a setting to toggle between default OS keybindings and app-specific shortcuts would be very helpful.
Yes, recent update made it so that Control-F - regardless of whether the cursor is focused on the text input field or not - pops and moves focus to the "Find" bar - even worse behavior than before, no respect to macOS standard navigation.
This "Find" behavior is also bound to Command-F, which I understand; but why bind it to Control-F? It is redundant, breaks norm, and is among the worst UI/UX behavior. Ask your in-house ChatGPT or anything if you doubt that. Plus, Codex is for coders who make extensive use of keybinding; and navigation shortcuts are something most of us are accustomed to. It does not make sense to have it behave in a non-standard manner.
So please, fix this.
@stym06
There are already at least 3 other duplicated issues created regarding this bug -- kinda making it obvious that the problem is annoying and affecting many mac users.
⌃-F is for moving cursor forward on macOS, period. Not to mention there is ⌘-F mapped for the same functionality (search) already.
Had the Desktop App been opensourced as well, a fix/PR would have been created days ago. 🙃
FYI. Just used codex to reverse engineering the Codex.app. It won't bother me anymore on my machine now. The resolution is like below, you may send it to your LLM.
@shengyang998 I just tried, and after two failed attempts, it gave up:
Yeah, it may have failed in the first several attempts. Here’s what I did. I copied the Codex.app to Codex Back.app. Then I opened the Back one and asked it to try to reverse engineer the original one. And I asked it to run until it found the way. Then I started to do something else. I saw it modify something and tried to open the original Codex app. I don’t know how many times it tried; it was just working on other things. And at the end, I found the Codex app just ran successfully. So I asked it to tell me what it did. And that’s what I texted here before.
I also sent this link to the Codex to help it understand the issue.
I don’t know why yours tells you it cannot be patched, but maybe let it try multiple times, or just let it loop until it is done. Hope it helps.
the Ctrl+F is a killer for Mac users, makes the CLI almost unusable if you are a Mac CLI user
The point is, ⌃-A, ⌃-E and ⌃-B still all work for moving cursor around, you would naturally assume ⌃-F do the same, and frequently press that out of habits and muscle memory.
Kudos to @ambrosino-oai to have given it a 'keyboard' label. Let's hope for a prompt fix (pun intended 🤣).
I just want to point out that this is a MAJOR issue for me too. It completely ruins my flow when I have to move my fingers off the home row and reach for the arrow keys just because of this. Thanks in advance for fixing this, @codex team.
I am with both Andy and Quincy - on a MAC ^F not mapping to forward char
and WORSE mapping to search is just plain broken and is killing my ability
to use the command line, for complex input i have to use emacs to edit and
then paste into the command area
John Schneider
C 408 203 7891
E @.***
On Fri, Apr 10, 2026 at 2:30 PM Quincy Larson @.***>
wrote:
Still reproducible for me on macOS in
26.409.20454 (1462).I initially thought a recent update might have fixed this. However, during this same update flow I also hit a separate in-app
updater issue: #17413.
Because of that updater failure, I’m not fully confident that the new version was applied cleanly, even though the app now reports
26.409.20454 (1462)locally.At least from what I can observe,
Ctrl+Fstill behaves incorrectly in the input field instead of doing the normal macOS forward-character movement.So either this issue is still present in
26.409.20454 (1462), or the updater bug in #17413 may have left the app in a partially updated/bad state.can confirm the issue still exists in 26.409.20454 (1462). The issue is marked as closed, but we may not assume the fix is shipped yet.
not fixed on Version 26.409.20454 (1462)
running MAC OS 26.3.1 (25D2128)
John Schneider
C 408 203 7891
E @.***
On Fri, Apr 10, 2026 at 10:31 PM Andy @.***> wrote:
It’s confusing that this is the earliest, most‑discussed report of the Ctrl+F bug and was marked “completed,” even though several people here have confirmed it’s still broken in recent versions (I can confirm that too!). Meanwhile, a newer duplicate (#17114) from last week with far fewer replies is left open. @etraut-openai
Anyway, I found a workaround for any Alfred user out there to try out. Please note that you have to pay for the Powerpack to be able to use workflows.
<img width="515" height="364" alt="Image" src="https://github.com/user-attachments/assets/2638833b-2ffd-40fa-99f5-c8c13340dfa8" />
Make sure the Hotkey is only registered in Codex by double-clicking the object, go to “Related Apps”, and drag Codex into the sheet. You can search for Codex in Alfred and drag the result from there.
<img width="646" height="528" alt="Image" src="https://github.com/user-attachments/assets/86d8fae6-071c-4e2a-aba1-2179ae178716" />
It's a bit laggy but it gets the job done.
@April-June-August So it sounds like I have to pay for a Powerpack add‑on just to work around a bug in this app?
@tranch I posted it here for the existing users who already paid.
To be clear, it’s not that this workflow (add-on) is paid, instead, once you unlock the Powerpack, you’ll be able to use every add-on there is or make your own ones.
And for the ones who have never heard of Alfred, I put the disclaimer that paid features are required so they won't go to download the app only to be frustrated.
So to answer your question: No, I don't think you have to pay for an app to work around a bug in Codex.
@openai Anyone taking a look at this?
If not, have one of your models take a look.
If you are somehow unable to do so, let me know - because I can fix it for you.
Clearly, it is a bug in Codex and no one should have to suffer implementing a workaround for this.
A viable, free workaround is to use Hammerspoon and bind ctrl-f to right key on 'Codex' app window only.
But that causes unnecessary delay (a few ms) for every control-F keystroke.
In short, OpenAI should fix this.
@ambrosino-oai Please reopen this bug.
This is an outstanding bug and nothing has been fixed.
@etraut-openai @dylan-hurd-oai @nornagon-openai @tibo-openai
Please, kindly, reopen this critical bug that still remains unresolved.
Just reproduced the issue on version 26.409.20454 (1462)
New version (26.415.20818; 1727) just arrived and I now confirm Control-F works as expected - the way it should (no find/search bar).
Many thanks to whoever made a commit for this.
Kudos to the Codex team. Love the browser and file viewer (CMD-P) as well.
Finally. After more than 3 months.