home / mcp / telegram mcp server
An MCP for telegram to integrate with Claude desktop.
Configuration
View docs{
"mcpServers": {
"alexandertsai-mcp-telegram": {
"command": "python",
"args": [
"/path/to/mcp-telegram/src/mcp_telegram/main.py"
],
"env": {
"TELEGRAM_API_ID": "YOUR_API_ID",
"TELEGRAM_API_HASH": "YOUR_API_HASH",
"TELEGRAM_SESSION_STRING": "SESSION_STRING_IF_USED"
}
}
}
}You can connect Claude to Telegram through this MCP server to read, manage, and respond to Telegram messages directly from your workflow. It provides chat listings, message history with read status, and context-aware replying to match your style.
After you have the MCP server running, you can interact with Claude to manage Telegram conversations. You can ask Claude to check unread chats, read message history from a specific chat, mark messages as read, or send messages to any chat. You can also request Claude to reply based on your conversation style for more natural interactions.
Use the MCP client to connect to the Telegram MCP server you set up. You will point the client at the local/remote server endpoint and then issue requests via Claude’s prompts. Practical usage includes:
- Check unread chats to see which conversations need attention.
- get_chats: List your Telegram chats with pagination, including chat names, IDs, and unread counts.
Once the server is configured and running, you can instruct Claude with natural language prompts to list chats, fetch messages, mark reads, or send messages to a chat.
List Telegram chats with IDs, names, and unread counts. Supports pagination; first page requires only the page number, subsequent pages use pagination parameters from the previous response.
Read messages from a specific chat with paginated history. Messages are automatically marked as read.
Mark all unread messages in a chat as read.
Send messages to any chat, with optional support for replying to a specific message.
Analyze your conversation style from convostyle.txt to help Claude respond in your preferred texting manner.