claude-code - 💡(How to fix) Fix [BUG] Auto-generated worktree/branch names contain ableist vocabulary (e.g. "crazy") [3 comments, 3 participants]

Preflight Checklist

• I have searched existing issues and this hasn't been reported yet
• This is a single bug report (please file separate reports for different bugs)
• I am using the latest version of Claude Code

What's Wrong?

Summary

Claude Code auto-generates worktree folders and branches using a Docker-style <adjective>-<surname>-<hash> scheme. The adjective pool contains ableist and otherwise offensive terms, most notably crazy. This is unacceptable for professional use and contradicts Anthropic's stated commitment to AI safety and ethics. Please remove these terms and provide opt-out mechanisms.

Concrete examples

Real names auto-generated in my environment:

• crazy-proskuriakova-a42862
• jovial-kilby-c8787a

The branch claude/crazy-proskuriakova-a42862 appears in the PR creation UI, commit history, CI logs, and on every team member's screen.

Why this is a problem

1. crazy is widely recognized as ableist language

Historically used as a slur against people with mental illness. Modern inclusive-language guides (Google, Microsoft, APA, Conscious Style Guide) explicitly recommend removing it from professional writing, UIs, and codebases. Other terms likely in the same word list — horny, naughty, sassy, etc. — carry similar sexist, demeaning, or otherwise problematic connotations.

2. Mechanical pairing with real surnames is a secondary insult

The surnames appear to be drawn from lists of real scientists and engineers (e.g. proskuriakova likely refers to Mayanist Tatiana Proskouriakoff). Prefixing such names with crazy- is disrespectful to the individuals and their legacy — particularly concerning given that these lists often foreground women and scientists of color.

3. The protest cost is shifted onto users

Developers are forced to see these strings daily. Anyone who finds them offensive must spend their own time filing issues, editing config, or renaming artifacts. This is the opposite of inclusive-by-default design.

4. An AI company normalizing slurs is especially serious

Anthropic publicly positions itself as a leader in AI safety and ethics. Shipping an official CLI that embeds slurs and displays them without user consent is a direct contradiction of that positioning and sets a harmful precedent.

5. Conflict with user domain

I am building i-harmony.ai, a Japanese AI companion app focused on users' emotional wellbeing (anxiety, loneliness, mental health). A development tool that renders ableist vocabulary on screen while I build a mental-health-adjacent product is grotesque. The tool's naming is actively at odds with my product's reason for existing.

Requests (in priority order)

1. Immediately remove slurs and offensive terms from the adjective list (crazy, horny, naughty, sassy, funky, etc.). Run the list through an inclusive-language linter such as alex-js.
2. Provide opt-out from the random-naming scheme entirely. e.g. settings.json key worktreeNaming: "timestamp" | "sequential" | "custom".
3. Support custom templates such as wt-{YYYYMMDD}-{shortId}.
4. Ship a bulk-rename command so existing users can clean up already-generated names.

References

• Google Developer Documentation Style Guide — Inclusive documentation
• Microsoft Writing Style Guide — Bias-free communication
• Conscious Style Guide — https://consciousstyleguide.com/
• alex-js — https://alexjs.com/

Thank you for taking this seriously.

What Should Happen?

In priority order:

1. Immediately remove slurs and offensive terms from the adjective list (crazy, horny, naughty, sassy, funky, etc.). Run the list through an inclusive-language linter such as alex-js.

2. Provide an opt-out from random naming entirely via settings.json, e.g.:

   { "worktreeNaming": "timestamp" }

with options like "timestamp" | "sequential" | "custom".

Support custom naming templates, e.g. wt-{YYYYMMDD}-{shortId} or {branch-name}. Ship a bulk-rename command so existing users can clean up already-generated names without manually invoking git worktree move and git branch -m. Consider dropping the "real-person surname" convention entirely. Neutral identifiers (hashes, timestamps, or branch-derived names) avoid the entire class of problems.

Steps to Reproduce

1. Install Claude Code (v2.1.113 or similar).
2. Run Claude Code in a git repository in a mode that creates a worktree
   (e.g. starting a new agent session with worktree isolation enabled).
3. Observe the auto-generated folder path under `.claude/worktrees/` and
   the corresponding branch name `claude/<adjective>-<surname>-<hash>`.
4. Repeat several times — the adjective pool will cycle through words
   including `crazy`, and pair them with real scientists' surnames.
Example output from my environment:
- `.claude/worktrees/crazy-proskuriakova-a42862` → branch `claude/crazy-proskuriakova-a42862`
- `.claude/worktrees/jovial-kilby-c8787a` → branch `claude/jovial-kilby-c8787a`

### Error Messages/Logs

```shell

Steps to Reproduce

Unknown / Not a regression as far as I can tell — this appears to be the designed behavior since worktree auto-naming was introduced. The request is to change the design, not to restore previous behavior.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

No response

Claude Code Version

2.1.113

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

https://i-harmony.ai/ 日本語で言わせてもらうが 君等のやっていることは一切許容できない。差別語をaiが内包し平然と使用するのは相当問題。何という下品なかつ不快な存在か 反省すべき。人名を付加するというのも全く理解できない 至急改善するべきです。i-harmony.ai 開発チーム