Encoding changing when editing EUC-KR files

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

It started occurring about 4–5 days ago.

When I edit a file containing Korean characters that is encoded in EUC-KR through the Agent, it is immediately converted to UTF-8.

All Korean text in the code becomes garbled, and even if I perform an Undo on the edit result, the encoding is not restored.

This happens regardless of the AI model being used, and it occurs consistently in all cases.

After that, Even if [Undo File], the encoding does not revert to the original.

Even if select [Undo File], the file has already been saved in UTF-8, so when reopened, it appears corrupted.

Even if [Reopen with Encoding], it still appears corrupted.

  • All MCP features from the cursor have been disabled at the cursor position.
  • It is in Agent mode.
  • The same occurs regardless of which model selected.
  • I’ve seen a few similar inquiries, but the solutions provided in those posts do not resolve the issue.

Attached Images are
#1. Before Using Agent
#2. After Using Agent
#3. After Using Agent and Undo File
#4. After Using Agent and Reopen with Encoding

Steps to Reproduce

Opening a file encoded in EUC-KR
Ask to Agent : change “Sweden” to “France”

Expected Behavior

  • The file encoding must not be changed.
  • Korean characters must not become corrupted.

Screenshots / Screen Recordings

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

+I updated the cursor a little while ago, and the same error keeps appearing. (2026-01-22, 5 PM KST)

Version: 2.4.20
VSCode Version: 1.105.1
Commit: 20b56586b0785b8843487045393c57c6d89b7100
Date: 2026-01-22T07:38:04.757Z
Build Type: Stable
Release Track: Early Access
Electron: 39.2.7
Chromium: 142.0.7444.235
Node.js: 22.21.1
V8: 14.2.231.21-electron.0
OS: Darwin arm64 25.2.0

For AI issues: which model did you use?

Composer, GPT 5.2 Sonnet 4.5

For AI issues: add Request ID with privacy disabled

Request ID : 8617561c-a96f-42b9-9fd4-5004aea51d8d

Does this stop you from using Cursor

Yes - Cursor is unusable

I’m experiencing the same issue with Windows-1252–encoded files. Any edit made by the agent automatically converts them to UTF-8.

Hey, thanks for the report.

This is a known encoding issue when editing via Agent. The team is aware. I added your report for EUC-KR files to increase visibility.

Related threads: Cursor returns UTF-8 code despite Windows-1252 configuration (VB6/Windows-1252), Cursor changes encoding of source code file (C#/Central European)

Temporary workaround:

  • After editing with Agent: bottom right corner of the editor → click the encoding → “Reopen with Encoding” → EUC-KR (to see the original)
  • Then “Save with Encoding” → EUC-KR
  • Or try CTRL+K (inline edit) instead of Agent. It might break the encoding less, but it’s not guaranteed.

@CarlosCarvalho same situation on your side. The workaround above should help for now.

Hey all!

Happy to report that this should be fixed when 2.5 comes out.

That seems too far off. I’ve been sticking with version 2.3.41 because of this issue and have had to give up on version 2.4 entirely. I’m really hoping version 2.5 gets released sooner rather than later.