Translator

Overview

This skill provides text translation services between six languages: Chinese (Simplified), English, Japanese, Korean, Russian, and Arabic. It supports two translation methods:

1. API-based translation (primary) - Uses external translation APIs for high-quality translation
2. AI-based translation (fallback) - Uses Claude's built-in translation capabilities when API is unavailable

Quick Start

Basic translation workflow:

User: "把'你好世界'翻译成英文"
→ Extract text: "你好世界"
→ Extract target language: "英文" (English)
→ Call translation script or use Claude
→ Return: "Hello World"

Core Workflow

Step 1: Identify Translation Request

Detect when user wants translation by looking for:
- Explicit triggers: "翻译", "translate", "翻译成", "translation"
- Implicit requests: "这段话用日语怎么说", "How do you say X in French"
- Language mentions: "中文转英文", "English to Japanese"

Step 2: Extract Text and Target Language

Parse the user's request to identify:
- Source text: The text to be translated
- Target language: The desired output language
- Source language (optional): If specified, use it; otherwise auto-detect

Examples:

Input: "把'你好世界'翻译成英文"
→ text: "你好世界"
→ target: "英文" (English)
→ source: (auto-detect = Chinese)

Input: "Translate 'Hello World' to Japanese"
→ text: "Hello World"
→ target: "Japanese"
→ source: (auto-detect = English)

Step 3: Perform Translation

Priority order:

1. Try API translation first (if configured)
   - Run scripts/translate.py with text, source, and target language
   - See API Setup for configuration

2. Fallback to Claude translation (if API fails or not configured)
   - Use Claude's built-in translation capability
   - Provide accurate translation directly

Step 4: Return Translated Text

Output the translated text in a clear format:

原文（中文）：你好世界
翻译（英文）：Hello World

Supported Languages

API Setup

To use API-based translation (higher quality, consistent results):

1. Get API credentials for one of these services:
   - Google Cloud Translation API
   - DeepL API
   - Microsoft Azure Translator
   - LibreTranslate (open-source, self-hosted)

2. Configure the script by setting environment variables:
   - GOOGLE_APPLICATION_CREDENTIALS for Google Cloud
   - DEEPL_API_KEY for DeepL
   - AZURE_TRANSLATOR_KEY for Azure
   - Or update scripts/translate.py with your API key

3. Test the setup by running:
   bash python scripts/translate.py --text "Hello" --target zh

If API is not configured or fails, the skill automatically falls back to Claude-based translation.

Usage Examples

Example 1: Simple Translation

User: "把'谢谢'翻译成英文"

Process:
1. Detect translation request
2. Extract: text="谢谢", target=English
3. Perform translation (API or Claude)
4. Return: "Thank you"

Example 2: Multi-line Text

User: "翻译成日文：
今天天气很好
我们出去玩吧"

Process:
1. Extract full text (preserve line breaks)
2. Target language: Japanese
3. Translate preserving formatting
4. Return:
今日は天気がとても良いです 私たちは遊びに出かけましょう

Example 3: Auto-detect Source Language

User: "Translate this to English: Bonjour le monde"

Process:
1. Extract text: "Bonjour le monde"
2. Auto-detect source: French
3. Target: English
4. Return: "Hello World"

Scripts

scripts/translate.py

Python script that performs API-based translation. Supports multiple translation services and fallback to Claude.

Usage:

python scripts/translate.py --text "Hello World" --target zh --source en

Parameters:
- --text: Text to translate (required)
- --target: Target language code (required, e.g., zh, en, ja)
- --source: Source language code (optional, auto-detect if omitted)
- --api: API service to use (optional, default: auto-select)

Supported APIs:
- google - Google Cloud Translation API
- deepl - DeepL API
- azure - Microsoft Azure Translator
- libre - LibreTranslate (open-source)

Notes

• This skill prioritizes simplicity and ease of use
• For advanced features (batch translation, document translation, preservation of formatting), consider extending this skill with additional scripts
• The translate.py script requires additional Python libraries depending on the API service chosen (e.g., google-cloud-translate, deepl, requests)
• If no API is configured, Claude's built-in translation capability provides a reliable fallback

References

No additional reference files needed. The skill is self-contained with the script and this SKILL.md.

Assets

No additional assets needed. Translation is performed via API or Claude directly.