Forked from NanoClaw. Stripped down to a single purpose: a Telegram bot that translates between Chinese and Japanese in real time.

Send a message to the Telegram bot:

• 中文 → 日本語 (Chinese to Japanese)
• 日本語 → 繁體中文 (Japanese to Traditional Chinese)

That's it. No commands, no menus. Every message gets translated and replied instantly.

Groq runs open-source LLMs (Llama 3.3 70B) on custom LPU hardware with extremely low latency. For a real-time translation bot, speed matters more than anything. Compared to other options:

• Groq — Free tier available, ~200ms inference, OpenAI-compatible API
• OpenAI / Anthropic — Higher quality but slower and paid-only
• Local models — No API costs but requires GPU hardware

Groq hits the sweet spot: fast enough for conversation flow, free to start, and the 70B model is more than capable for translation tasks.

git clone https://github.com/your-username/cj-bot.git
cd cj-bot
npm install
cp .env.example .env

Edit .env with your credentials:

TELEGRAM_BOT_TOKEN=your-telegram-bot-token
GROQ_API_KEY=your-groq-api-key
ALLOWED_TELEGRAM_IDS=your-telegram-user-id

Run:

npm run dev

1. Open Telegram, search for @BotFather
2. Send /newbot
3. Enter a display name for your bot (e.g. CJ Translator)
4. Enter a username for your bot (must end in bot, e.g. cj_translator_bot)
5. BotFather will reply with your bot token, looks like 123456789:ABCdefGhIJKlmNoPQRsTUVwxyZ
6. Copy the token into .env as TELEGRAM_BOT_TOKEN

1. Go to console.groq.com and sign up (free)
2. In the left sidebar, click API Keys
3. Click Create API Key, give it a name
4. Copy the generated key (starts with gsk_) into .env as GROQ_API_KEY

Only whitelisted users can use the bot. To find your Telegram user ID:

1. Open Telegram, search for @userinfobot
2. Send any message — it will reply with your user ID (a number like 123456789)
3. Copy the ID into .env as ALLOWED_TELEGRAM_IDS

Multiple user IDs can be comma-separated: 123,456,789

• Only whitelisted Telegram user IDs can use the bot
• Messages sent before bot startup are ignored
• Prompt injection defenses built into the system prompt
• API keys stored in .env (excluded from git)

MIT