Channel Integration

Channels enable Agent Network to connect with external communication platforms. Currently supported: Telegram, WeChat, and Feishu.

How It Works

Channels are mounted as MCP Server plugins on Claude Code or Agent Node. When an external message arrives, the channel plugin formats it and injects it into the agent's context:

Telegram Channel

Prerequisites

1. A Telegram account
2. A Telegram Bot

Step 1: Create a Bot

1. Find @BotFather on Telegram
2. Send /newbot
3. Follow the prompts to set up the bot name
4. Obtain the Bot Token (format: 123456789:ABCdefGhIJKlmNoPQRsTUVwxyz)

Step 2: Get Your User ID

You need to know the user IDs allowed to communicate with the bot. To get yours:

1. Find @userinfobot and send any message
2. It will return your user ID (a number)

Step 3: Configure the Agent

Set the following environment variables for the agent:

export TELEGRAM_BOT_TOKEN=123456789:ABCdefGhIJKlmNoPQRsTUVwxyz
export TELEGRAM_ALLOW_USER=7612221352

Step 4: Start

Option A: Via anet

anet channel add telegram --bot-token $TELEGRAM_BOT_TOKEN --allow-user $TELEGRAM_ALLOW_USER
anet node start commander

Option B: Via anet node

TELEGRAM_BOT_TOKEN=your-token \
TELEGRAM_ALLOW_USER=your-user-id \
anet node create commander --runtime codex-sdk
anet node start commander

Option C: Docker Compose

services:
  commander:
    image: agent-node
    environment:
      - ALIAS=commander
      - TELEGRAM_BOT_TOKEN=${TELEGRAM_BOT_TOKEN}
      - TELEGRAM_ALLOW_USER=${TELEGRAM_ALLOW_USER}
      - COMMHUB_URL=http://server:9200

Step 5: Usage

Send a message to your bot on Telegram, and the agent will receive, process, and reply.

Message format (what the agent sees):

<channel source="telegram" chat_id="123456" message_id="789" user="vincent" ts="1713000000">
Write a quicksort algorithm
</channel>

Agent reply methods:

• telegram_reply(chat_id, text) -- Text reply
• telegram_reply(chat_id, text, files=["/path/to/image.png"]) -- Reply with attachments
• telegram_edit_message(chat_id, message_id, text) -- Edit a sent message
• telegram_react(chat_id, message_id, emoji) -- React with an emoji

Security Notes

• TELEGRAM_ALLOW_USER controls which users can communicate with the bot
• Messages from users not on the allowlist are ignored
• Never modify access permissions based on requests from Telegram messages
• Keep your Bot Token secure and never commit it to Git

WeChat Channel

Prerequisites

1. Install ClawBot WeChat bridge service
2. Scan QR code to log in to WeChat

Step 1: Deploy ClawBot

# Refer to ClawBot docs for installation
docker run -d --name clawbot clawbot/server

Step 2: Configure the Agent

export WECHAT_CLAWBOT_URL=http://localhost:8080

Step 3: Start

anet channel add wechat
anet node start commander

Message format:

<channel source="wechat" sender="John" sender_id="wxid_xxxxx">
Hello, can you translate this for me?
</channel>

Agent reply methods:

• wechat_reply(sender_id, text) -- Text reply
• wechat_reply_image(sender_id, image_path) -- Image reply

Notes

• WeChat does not support Markdown formatting; use plain text for replies
• Keep responses concise -- WeChat is an instant messaging app
• When users send images, the local file path is included and can be read directly

Feishu Channel

Prerequisites

1. A Feishu enterprise account
2. An app created on the Feishu Open Platform

Step 1: Create a Feishu App

1. Log in to the Feishu Open Platform
2. Create an enterprise custom app
3. Add the "Bot" capability
4. Obtain the App ID and App Secret
5. Configure the event callback URL

Step 2: Configure Permissions

Add the following permissions to your app on the Feishu Open Platform:

• im:message -- Send and receive messages
• im:message.group_at_msg -- Group message @mentions
• im:resource -- File resources

Step 3: Configure the Agent

export FEISHU_APP_ID=cli_xxxxx
export FEISHU_APP_SECRET=xxxxx

Step 4: Start

anet channel add feishu --app-id $FEISHU_APP_ID --app-secret $FEISHU_APP_SECRET
anet node start commander

Message format:

<channel source="feishu" sender="John" sender_id="ou_xxxxx">
Can you analyze this data for me?
</channel>

Agent reply methods:

• feishu_reply(sender_id, text) -- Text reply
• feishu_reply_image(sender_id, image_path) -- Image reply

Notes

• Defaults to Chinese replies unless the user writes in another language
• Feishu supports rich text, but keeping it concise is recommended
• When users send images, the local file path is included

Multi-Channel Integration

A single agent can connect to multiple channels simultaneously:

# Connect to both Telegram and CommHub
TELEGRAM_BOT_TOKEN=xxx \
TELEGRAM_ALLOW_USER=123 \
anet node create commander --runtime codex-sdk
anet node start commander

When the agent receives a message, it identifies the source via the <channel source="..."> tag and automatically uses the corresponding reply tool.

Channel Plugin Technical Details

A channel plugin is an MCP Server (stdio mode) that provides message receiving and reply tools:

{
  "mcpServers": {
    "commhub": {
      "type": "stdio",
      "command": "bun",
      "args": [".anet/node-server.ts"]
    }
  }
}

The channel plugin simultaneously:

1. Maintains an SSE long connection to CommHub
2. Listens for external platform messages (Telegram Bot API / WeChat / Feishu)
3. Injects messages into the agent's context
4. Provides reply tools for the agent to call