[MODEL] Claude Code can not stop using the word "load-bearing"
Preflight Checklist
- [x] I have searched existing issues for similar behavior reports
- [x] This report does NOT contain sensitive information (API keys, passwords, etc.)
Type of Behavior Issue
Other unexpected behavior
What You Asked Claude to Do
I'm using Claude Code to build software. I'm mostly using the superpowers plugin to structure my workflow. There is not one specific thing I prompted that is causing the behaviour.
I'm using Opus 4.7 (1M context), max effort.
What Claude Actually Did
Since a few weeks, Claude Code very frequently uses the word "load-bearing" in chat output, documents it writes, commit messages, and so on. It never used to do that. It's completely overused and extremely ugly.
I searched for the word "load-bearing" in all plugins, CLAUDE.md files, and any other prompts it may have read and there is no mention of the word. I have told it to save to memory to NOT use the word "load-bearing" but it can't help itself.
Expected Behavior
I expect Claude to not overuse the word "load-bearing" and not see it every 10 paragraphs.
Files Affected
Permission Mode
Accept Edits was ON (auto-accepting changes)
Can You Reproduce This?
Yes, just run a claude code session for an hour and it'll probably say "load-bearing" at least once.
Steps to Reproduce
_No response_
Claude Model
Opus 4.7
Relevant Conversation
Impact
Minor
Claude Code Version
2.1.120 (Claude Code)
Platform
Anthropic API
Additional Context
_No response_
This issue has 11 comments on GitHub. Read the full discussion on GitHub ↗