reply-language
A Claude Code hook that keeps replies in Traditional Chinese (Taiwan)
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/annihilationwizard--reply-language)HTML
<a href="/listings/annihilationwizard--reply-language"><img src="/api/listings/annihilationwizard--reply-language/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
In a Claude Code **terminal** session, run:
/plugin install reply-language --marketplace AnnihilationWizard/reply-language
1. Answer `y` to add the `github:AnnihilationWizard/reply-language` marketplace. 2. Choose a scope. The **user** scope runs the hook in every session. 3. Claude Code confirms `Plugin is now active.` No restart is needed.
It is also listed in the [AnnihilationWizard/claude-hooks](https://github.com/AnnihilationWizard/claude-hooks) marketplace as `reply-language@claude-hooks`.
> [!TIP] > If the same script is also registered as a Stop hook in `~/.claude/settings.json`, remove that entry after installing, or each reply is checked twice.
To update later, run `claude plugin update`, then `/reload-plugins`.
<details> <summary>Run from a local clone instead</summary>
git clone https://github.com/AnnihilationWizard/reply-language.git claude --plugin-dir ./reply-language
To load a clone in every session, add it to the `env` block of `~/.claude/settings.json`:
{
"env": { "CLAUDE_CODE_PLUGIN_DIRS": "/path/to/reply-language" }
}</details>
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
.claude-plugin/marketplace.json · Manifest · 9 lines
- {
- "name": "reply-language",
- "description": "A Claude Code hook that keeps replies in Traditional Chinese (Taiwan)",
- "owner": {
- "name": "AnnihilationWizard",
- "url": "https://github.com/AnnihilationWizard"
- },
- "plugins": [{ "name": "reply-language", "source": "./" }]
- }
README.md · README · 131 lines
- # reply-language
- <p align="center">
- <img src="docs/images/icon.svg" width="96" height="96" alt="reply-language icon: a speech bubble holding 中 with a check mark">
- </p>
- <p align="center">
- <strong>Replies stay in Traditional Chinese.</strong><br>
- A Claude Code Stop hook that sends a reply written mostly in English back to Claude, to rewrite in Traditional Chinese with Taiwan terminology.
- </p>
- <p align="center">
- <a href="CHANGELOG.md"><img src="https://img.shields.io/badge/Version-1.0.0-green.svg" alt="Version 1.0.0"></a>
- <a href="LICENSE"><img src="https://img.shields.io/badge/License-MIT-blue.svg" alt="MIT License"></a>
- <a href="https://code.claude.com/docs"><img src="https://img.shields.io/badge/Platform-Claude%20Code-D97757.svg" alt="Platform: Claude Code"></a>
- <a href="hooks/hooks.json"><img src="https://img.shields.io/badge/Type-Stop%20hook-6E7681.svg" alt="Stop hook"></a>
- <a href="hooks/reply-language-check.test.mjs"><img src="https://img.shields.io/badge/Tests-5%20passing-3FB950.svg" alt="5 tests passing"></a>
- </p>
- <p align="center">
- <a href="#features">Features</a> ·
- <a href="#install">Install</a> ·
- <a href="#usage">Usage</a> ·
- <a href="#limitations">Limitations</a> ·
- <a href="#development">Development</a> ·
- <a href="CONTRIBUTING.md">Contributing</a> ·
- <a href="CHANGELOG.md">Changelog</a>
- </p>
- 
- *Illustration of a session before and after.*
- Instructions such as "reply in Traditional Chinese" in `CLAUDE.md` work most of the time, but after a long run of tool calls Claude can slip back into English. reply-language checks each reply when Claude ends its turn. When the reply is mostly English, Claude gets it back once with the instruction to rewrite the same reply in Traditional Chinese, keeping code, commands, paths and identifiers as they are.
- ## Features
- - **Checks the final reply only:** The hook reads the transcript back to the last user-role entry: a prompt, a tool result or a subagent hand-back. English status lines written between tool calls earlier in the turn do not count.
- - **Technical terms are free:** Code blocks, inline code, URLs and paths are removed before counting, so a Chinese reply full of `git` commands and file names passes.
- - **One rewrite per reply:** When Claude is already rewriting because of this hook, the reply is not checked again, so a turn can never loop.
- - **Short replies pass:** A reply with fewer than 40 letters and Han characters is never sent back.
- - **Nothing to configure:** One Node.js script with no dependencies, registered as a Stop hook by the plugin.
- | Final reply | Result |
- | --- | --- |
- | Mostly Chinese, with commands in backticks or code blocks | Passes |
- | Mostly Chinese, after English status lines earlier in the turn | Passes |
- | Mostly English, at least 40 letters and Han characters | Sent back once to rewrite |
- | Shorter than 40 letters and Han characters | Passes |
- ## Install
- In a Claude Code **terminal** session, run:
- ```
- /plugin install reply-language --marketplace AnnihilationWizard/reply-language
- ```
- 1. Answer `y` to add the `github:AnnihilationWizard/reply-language` marketplace.
- 2. Choose a scope. The **user** scope runs the hook in every session.
- 3. Claude Code confirms `Plugin is now active.` No restart is needed.
- It is also listed in the [AnnihilationWizard/claude-hooks](https://github.com/AnnihilationWizard/claude-hooks) marketplace as `reply-language@claude-hooks`.
- > [!TIP]
- > If the same script is also registered as a Stop hook in `~/.claude/settings.json`, remove that entry after installing, or each reply is checked twice.
- To update later, run `claude plugin update`, then `/reload-plugins`.
- <details>
- <summary>Run from a local clone instead</summary>
- ```bash
- git clone https://github.com/AnnihilationWizard/reply-language.git
- claude --plugin-dir ./reply-language
- ```
- To load a clone in every session, add it to the `env` block of `~/.claude/settings.json`:
- ```json
- {
- "env": { "CLAUDE_CODE_PLUGIN_DIRS": "/path/to/reply-language" }
- }
- ```
- </details>
- ## Usage
- Once installed, there is nothing to configure. When a reply is mostly English, Claude reads:
- ```text
- The reply to the user is mostly in English. Rewrite the same reply in Traditional Chinese
- with Taiwan terminology, keeping code, commands, paths and identifiers as they are.
- ```
- ### What counts as English
- After code blocks, inline code, `http` and `https` URLs, and paths with at least one `/` are removed, the hook counts Latin letters and Han characters in what is left. The reply is English when the two together reach 40 and there are more than twice as many Latin letters as Han characters.
- ## Limitations
- - Plugins do not load in Claude Code on the web (cloud sessions). To use the hook there, commit a copy of the script to the repository and register it in that repository's `.claude/settings.json`.
- - Only the language mix is checked, not the terminology. A reply in Simplified Chinese, or with mainland China terms, passes.
- - A reply that is mostly English prose outside code formatting, such as a quoted error message, is sent back once. Put such text in a code block.
- - The thresholds are fixed in the script.
- ## Development
- ```bash
- claude plugin validate --strict . # manifest, marketplace file and hooks
- node --test hooks/ # runs the hook on sample transcripts
- ```
- Run `claude --plugin-dir .` to try changes. A change reaches installed copies only when `version` in `.claude-plugin/plugin.json` goes up.
- ## Project Structure
- ```text
- .claude-plugin/ plugin.json manifest and marketplace.json
- hooks/ hooks.json, the Stop hook script and its tests
- docs/images/ README icon and illustration
- ```
- ## Contributing
- Bug reports and ideas are welcome. Read [CONTRIBUTING.md](CONTRIBUTING.md) before opening an [issue](https://github.com/AnnihilationWizard/reply-language/issues) or pull request, and follow [SECURITY.md](SECURITY.md) for security reports.
- ## License
- [MIT License](LICENSE). Copyright © 2026 [AnnihilationWizard](https://github.com/AnnihilationWizard).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.