Telegram

Telegram is the recommended channel. It's easy to set up and supports voice transcription.

Setup

1. Create a Bot

• Open Telegram and search for @BotFather
• Send /newbot, follow the prompts
• Copy the bot token

2. Get Your User ID

• Message @userinfobot on Telegram
• Copy your User ID

3. Configure

{
  "channels": {
    "telegram": {
      "enabled": true,
      "token": "YOUR_BOT_TOKEN",
      "allow_from": ["YOUR_USER_ID"]
    }
  }
}

Field	Type	Description
enabled	bool	Enable/disable the channel
token	string	Bot token from @BotFather
base_url	string	Custom Telegram Bot API server URL (optional)
proxy	string	HTTP/SOCKS proxy URL (optional, also reads HTTP_PROXY env)
allow_from	array	List of allowed user IDs (empty = allow all)
reasoning_channel_id	string	Route reasoning output to a separate chat
group_trigger	object	Group chat trigger settings (mention_only, prefixes)

4. Run

picoclaw gateway

Voice Transcription

Telegram voice messages can be automatically transcribed using Groq's Whisper:

{
  "model_list": [
    {
      "model_name": "whisper",
      "model": "groq/whisper-large-v3",
      "api_key": "gsk_..."
    }
  ]
}

Get a free Groq API key at console.groq.com.

Troubleshooting

"Conflict: terminated by other getUpdates": Only one picoclaw gateway can run at a time. Stop any other instances.

Proxy: If Telegram is blocked in your region, use the proxy field:

{
  "channels": {
    "telegram": {
      "proxy": "socks5://127.0.0.1:1080"
    }
  }
}

Bot Commands

The Telegram channel registers these built-in bot commands:

Command	Description
/start	Greeting message
/help	Show help text
/show [model|channel]	Show current configuration
/list [models|channels]	List available options

Quoted Reply Support

When a user replies to a message in Telegram (using the built-in "Reply" feature), PicoClaw automatically includes the quoted message as context for the agent. This works for both user messages and the bot's own previous responses:

• Text: The quoted message text is prepended to the user's new message as [quoted user/assistant message from author]: ...
• Media: If the quoted message contains voice or audio, those files are also downloaded and included in the agent's input
• Role detection: The bot distinguishes whether the quoted message was from a user, the bot itself (assistant), or another bot

This allows the agent to understand conversational context even when messages are not sent consecutively.

Media Support

The bot handles photos, audio files, documents, and voice messages. Voice messages are transcribed if a Whisper model is configured (see Voice Transcription).