Copying plain text from assistant output inserts unwanted line breaks
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
When the assistant outputs plain text (e.g., a paragraph
for the user to copy/paste into LinkedIn, email, etc.),
copying the rendered text from the terminal inserts
hard line breaks at the terminal's column width. A
single paragraph becomes ~14 lines with breaks at
arbitrary points. This happens regardless of whether the
text uses markdown formatting or not.
The only workaround is having the assistant write the
text to a temp file and then running ! cat /tmp/file.txt
to copy from raw shell output instead.
What Should Happen?
Copying plain text from the assistant's rendered output
should preserve the original paragraph structure — a
single paragraph should paste as a single paragraph
without inserted line breaks.
Error Messages/Logs
Steps to Reproduce
- Open Claude Code in a terminal (tested at ~100 column
width)
- Ask: "Write me a 3-sentence LinkedIn comment about
real estate technology"
- Claude outputs the text
- Select and copy the output text
- Paste into any text field (browser, Notes, etc.)
- Result: line breaks appear every ~80-100 characters,
breaking the paragraph into multiple lines
Claude Model
Opus
Is this a regression?
Yes, this worked in a previous version
Last Working Version
Not sure — may have always been this way, but it's particularly noticeable when using Claude Code for drafting copy/messages.
Claude Code Version
2.1.100
Platform
Anthropic API
Operating System
Other Linux
Terminal/Shell
Other
Additional Information
Pinned to v2.1.100 — VPS CPU lacks AVX instruction set,
newer versions crash with illegal opcode. Bug is likely
reproducible on latest version as it's a terminal
rendering issue, not version-specific.
OS: AlmaLinux 8.10 (RHEL-based), kernel 4.18.0-553.
Terminal: Bash over SSH (HTTPS terminal session). Claude
Code pinned to v2.1.100 (VPS CPU lacks AVX, newer
versions crash).
The workaround of writing to
/tmp/reply.txt and using ! cat /tmp/reply.txt works,
which confirms the issue is in the terminal rendering
layer, not the text content itself. The assistant's raw
output has no hard line breaks — they're introduced
during rendering.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗