Skip to content

IME input issue on Windows: first character is not captured correctly when typing without focus #18930

@yuuuuuuuu25

Description

@yuuuuuuuu25

What version of the Codex App are you using (From “About Codex” dialog)?

26.415

What subscription do you have?

Business

What platform is your computer?

Microsoft Windows 11 x64

What issue are you seeing?

The Codex App supports typing even when the input box is not focused (similar to ChatGPT). However, this behavior does not work correctly with Japanese IME input.

When typing Japanese text using IME (e.g. typing "さくら"), the expected input sequence is:
s → a → k → u → r → a

However, when starting to type without focusing the input box first, the first character is not correctly handled by the IME.

For example:
Typing "sakura" results in:
sあくら

The first character ("s") is inserted as a raw character instead of being processed by the IME.

Steps to reproduce:

  1. Launch Codex App on Windows
  2. Make sure the text input field is NOT focused
  3. Start typing Japanese using IME (e.g. "sakura")
  4. Observe the resulting text

Expected behavior:

The input should be handled by the IME from the first keystroke, resulting in proper Japanese text (e.g. "さくら")

Actual behavior:

The first character is inserted as a raw Latin character, and the rest is processed by IME

Additional notes:

This seems to be related to how the app handles focus and key events before IME composition starts.

IME:

Microsoft IME (default Windows Japanese IME)

Reproducibility:

100%

2026-04-22.104548.mp4

What steps can reproduce the bug?

  1. Launch Codex App on Windows
  2. Make sure the text input field is NOT focused
  3. Start typing Japanese using IME (e.g. type "sakura")
  4. Observe the resulting text

What is the expected behavior?

The input should be handled by the IME from the first keystroke, resulting in proper Japanese text (e.g. "さくら").

Typing "sakura" should correctly produce "さくら" even when starting to type without focusing the input field.

Additional information

This issue also occurs in ChatGPT with the same behavior when typing without focusing the input field.

When typing Japanese using IME without focusing the input field first, the first character is inserted as a raw Latin character.

For example:
Typing "sakura" results in:
sあくら

This suggests that the first key event is being processed before IME composition starts.

IME: Microsoft IME (default Windows Japanese IME)
Reproducibility: 100%

Metadata

Metadata

Assignees

No one assigned

    Labels

    appIssues related to the Codex desktop appbugSomething isn't workingwindows-osIssues related to Codex on Windows systems

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions