title: Telegram description: Connect a Telegram Bot to HiTechClaw for receiving and sending messages.

Prerequisites

• A Telegram Bot token from @BotFather

Setup

1. Create a Bot

1. Open Telegram and search for @BotFather
2. Send /newbot and follow the prompts
3. Copy the bot token (e.g., 123456789:ABCdefGHIjklMNOpqrSTUvwxYZ)

2. Configure in HiTechClaw

1. Navigate to Channels → Add Channel
2. Select Telegram
3. Paste your Bot Token
4. Click Test Connection — you should see your bot's username
5. Click Create to save

3. Activate

Click Activate on the channel card. HiTechClaw will start polling for messages.

How It Works

• Uses long polling (getUpdates API) to receive messages
• Group support: Bot responds when @mentioned, replied to, or when a command is sent
• Private chats: All messages are processed
• Typing indicator: Shows "typing..." while processing
• Message splitting: Long responses are automatically split at 4096 chars

Configuration Fields

| Field | Required | Description | |-------|----------|-------------| | botToken | ✅ | Bot token from @BotFather |

Image & Vision Support

When a user sends a photo, HiTechClaw automatically:

1. Downloads the image from Telegram's file server
2. Converts it to a base64 data URL
3. Hard-switches to the vision model (e.g. llava:13b via Ollama, or Gemini) for that message
4. The vision fallback chain is: ollama-vision → gemini → openai → ollama

Configure the vision model via env vars:

VISION_PROVIDER=ollama
VISION_MODEL=llava:13b   # or qwen2.5vl:7b, gemma3:12b

No config is needed if LLM_MODEL already supports vision (e.g. gemma3:12b).

Realtime Search Detection

For messages about time-sensitive topics (weather, news, prices, scores), HiTechClaw automatically prepends a routing instruction telling the agent to use web-search tools first. This works for both Vietnamese and English.

Group Chat

In group chats, the bot responds when:

• @mentioned by username
• Replied to directly
• A /command is sent